Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberCN1204754 C
Publication typeGrant
Application numberCN 02802661
PCT numberPCT/JP2002/008129
Publication date1 Jun 2005
Filing date8 Aug 2002
Priority date15 Aug 2001
Also published asCN1465194A, DE60239408D1, EP1421795A1, EP1421795B1, US6733138, US20030052837, WO2003017679A1
Publication number02802661.6, CN 02802661, CN 1204754 C, CN 1204754C, CN-C-1204754, CN02802661, CN02802661.6, CN1204754 C, CN1204754C, PCT/2002/8129, PCT/JP/2/008129, PCT/JP/2/08129, PCT/JP/2002/008129, PCT/JP/2002/08129, PCT/JP2/008129, PCT/JP2/08129, PCT/JP2002/008129, PCT/JP2002/08129, PCT/JP2002008129, PCT/JP200208129, PCT/JP2008129, PCT/JP208129
InventorsR拉斯卡尔
Applicant三菱电机株式会社
Export CitationBiBTeX, EndNote, RefMan
External Links: SIPO, Espacenet
Method and system for forming mosaic using multiple projectors
CN 1204754 C
Abstract  translated from Chinese
本发明公开了一种用多个投影机在显示表面上形成镶嵌图像的方法。 The present invention discloses a method of forming a mosaic image on a display surface with a plurality of projectors method. 由于各投影机在显示表面上依次投影对齐图像,使其被投影的对齐图像的组合体形成多边形。 As a result of the projector is projected on the display surface in order to align the image, making it a combination of the projected image is aligned forming polygon. 另外,用照相机为各对齐图像依次取得对应的输入图像。 In addition, for the alignment of the camera with the image corresponding to the input image sequentially acquired. 然后,特别指定由多边形所围成的显示表面上的显示区域,并为各投影机确定显示区域和各输入图像之间的一个投影矩阵。 Then, specifying the display region surrounded by the polygon display surface, and a projection matrix region and between each of the input image is determined for the projector display. 各投影机的源图像按对应投影机的单对应性而被翘曲。 The source of each projector's image by the homography of the corresponding projector is warped. 该翘曲的源图像的像素按一个投影矩阵进行加权,然后,在显示表面上直接被翘曲,与此同时投影其加权的源图像,从而形成镶嵌图像。 The warped source image pixels are weighted according to a projection matrix, then, warping directly on the display surface, while projected weighted source images, thereby forming a mosaic image.
Claims(10)  translated from Chinese
1.一种用多个投影机在显示表面上形成镶嵌图像的方法,其特征在于包括:为所述各投影机依次在所述显示表面上投射配准图像,使该被投射的配准图像的组合体形成多边形的步骤;为所述各投射到所述显示表面的配准图像依次用照相机取得对应的输入图像的步骤;识别由所述多边形围成的所述显示表面上的显示区域的步骤;通过利用从所述输入图象中提取出来的单个的配准图像,在照相机和投影机之间为每个投影机定义一个第一单对应性的步骤;通过利用投射到显示区域以外并重叠到其他投射的图像的部分配准图像,在显示区域和照相机之间为每个投影机定义一个第二单对应性的步骤;通过将所述第一和第二单对应性组合在一起,为所述各投影机确定所述显示区域和所述各输入图像之间的单个的投影矩阵的步骤;为所述各投影机把源图像按照所述对应的一个投影矩阵翘曲的步骤;按照所述一个投影矩阵把所述翘曲的源图像的像素加权的步骤;以及直接在所述显示表面上同时投射所述翘曲的加权的源图像,并形成所述镶嵌图像的步骤。 A method of using a plurality of projectors in the mosaic image formed on the surface of the display, characterized by comprising: sequentially to each of said projectors projecting said display image on the surface of registration, so that the projected image registration polygon forming step of the assembly; each of said projected onto the display surface of the image registration step of sequentially corresponding to the input image obtained by a camera; on the surface of the display region indicated by the identification of the polygon enclosed in By using the extracted from the input image out of a single registration images between the camera and projector in a first homography defined steps for each projector;; step by using the projected outside the display area and overlapping portion allocated to other registered images projected image between the display area and the camera of a second homography defined steps for each projector; by the first and second one correspondence together, each of the projector and the step of determining a single region of each of the input image between the display projection matrix; each of the source image to the projector according to the projection matrix corresponding to a warp step; according The projection matrix to the one step warped source image pixel weighting; and a step directly on the display while the image projected on the surface of the source the weighted warp and form the image mosaic.
2.如权利要求1所述的方法,其特征在于:所述显示表面对所述投影机中的至少一个光轴倾斜。 2. The method according to claim 1, wherein: said display surface of said at least one optical axis of the projector is inclined.
3.如权利要求1所述的方法,其特征在于:一个所述投影机的光轴至少对一个其它所述投影机的光轴倾斜。 The method according to claim, characterized in that: an optical axis of the projector to the optical axis of said at least one other of the tilt of the projector.
4.如权利要求1所述的方法,其特征在于:所述各投影机的光轴对所有其它的所述投影机的光轴倾斜。 4. The method according to claim 1, wherein: the optical axis of each projector is inclined to the optical axis of all other projector.
5.如权利要求1所述的方法,其特征在于:所述配准图像含有棋盘图形。 5. The method according to claim 1, characterized in that: the registration image comprising a checkerboard pattern.
6.如权利要求1所述的方法,其特征在于:所述显示区域是由所述多边形围成的最大可能的长方形。 6. The method according to claim 1, characterized in that: the region is surrounded by a polygon as a rectangle the largest possible display.
7.如权利要求1所述的方法,其特征在于:所述第一单对应性Hi是最大似然估计问题,还包含:最小费用函数Σn||u-Hix||]]>式中n是表示所述输入图像x和所述配准图像u之间的对应特征的数量。 N || u-Hix ||]]> type; the minimum cost function & Sigma: 7. The method according to claim 1, characterized in that: the first homography Hi is the maximum likelihood estimation, further comprising where n is a number corresponding to the characteristics of the image and the registered image x between the input u.
8.如权利要求7所述的方法,其特征在于:两个任意的源图像间的像素的映射为uj~=HjHi-1ui,式中,ui与uj表示所述源图像中的对应像素,Hj为所述第二单对应性。 8. The method according to claim 7, wherein: pixel mapping between two arbitrary source image is uj ~ = HjHi-1ui, where, ui and uj representing the source image corresponding pixel, Hj is the second single correspondence.
9.如权利要求1所述的方法,其特征在于:还包括当所述特定的像素在所述显示区域的外侧时,在各翘曲的图像中所述特定的像素上分配0权重的步骤;当所述特定的像素成为照明所述显示区域的唯一的像素时,在所述特定的像素上分配1权重的步骤;在以上情况以外时,在所述特定的像素上分配0<W<1范围的权重W的步骤,W和到具有0权重的最接近像素间的距离成比例。 9. The method according to claim 1, characterized in further comprising: when said particular pixel in said outer region of the display, the step of assigning a weight of 0 on each specific pixel in the warped image ; when illuminating said specific pixel becomes the pixel display region only, a weight in the step of allocating a specific pixel; in the case other than the above assigned 0 <W on the particular pixel < Step right range of weight W 1, the distance is proportional to W and to have zero weight closest pixels.
10.一种用以在显示表面上形成镶嵌图像的系统,其中包括:多个投影机,每个投影机用来将配准图像投射到所述显示表面上,使所述被投射的配准图像的组合体形成多边形;照相机,所述照相机用来为所述投射到所述显示表面的各配准图像依次取得对应的输入图像;第一定义单元,用来通过利用从所述输入图象中提取出来的单个的配准图像,在照相机和投影机之间为每个投影机定义一个第一单应性;第二定义单元,用来通过利用投射到显示区域以外并重叠到其他投射的图像的部分配准图像,在显示区域和照相机之间为每个投影机定义一个第二单对应性;识别单元,用以识别由所述多边形围成的所述显示表面上的显示区域;确定单元,用来为所述各投影机,在所述显示区域和所述各输入图像之间确定单独的投影矩阵;翘曲单元,用来为所述各投影机使源图像按照所述对应的单独的投影矩阵翘曲;以及加权装置,用来按照所述单独的投影矩阵,对翘曲的所述各源图像的像素加权,并把所述投影的、翘曲的和加权的源图像在所述显示表面上形成所述镶嵌图像。 10. A method for forming the mosaic image on the display surface of the system, including: multiple projectors, each projector for the registration image is projected onto the display surface, so that the projected registration image forming polygon assembly; camera, the camera used to display the projected onto the surface of each of the registration image corresponding to the input image acquired sequentially; a first defined cell for the input image by using the extracted from a single image registration between a first camera and projector homography is defined for each projector; second defined cell for the display area by using the projected outside and to overlap the other projected assigned reference image portion of the image between the display area and each camera defines a second projector homography; identification unit for identifying a polygon surrounded by said display area on said surface; determining unit for each of the projectors in the display area defined between each of the input image and the projection matrix alone; warp unit, used to make the various projector in accordance with the corresponding source image individual warp projection matrix; and weighting means for projection matrix according to the individual, to warp the image pixel weighting each source and said projection, warped and weighted source images The display of the mosaic image is formed on the surface.
Description  translated from Chinese
利用多个投影机在显示表面上形成镶嵌图形的方法和系统 Method and system for utilizing multiple projectors on the display surface mosaic pattern

技术领域 Technical Field

本发明涉及一种在平坦的表面上投影图像的光学装置,更具体地说涉及一种为形成一个镶嵌显示图像,投影任意个图像的方法。 The present invention relates to an optical device on a flat surface projection image, and more particularly to a display to form a mosaic image, images projected in any way.

背景技术 Background

照片镶嵌块是为了形成一张可分辨大的图像而相互配准的二维排列的图像。 Photo mosaic block in order to form a large image can be distinguished from each other with a two-dimensional array of image registration. 这些图像或是从同一观察点在各个方向拍摄的,或是从平面景物的各种观察点拍摄的。 These images are from the same observation point or shooting in all directions, or from various observation points in the plane of the shooting scene. 可参考如修切利斯基(Szeliski)等公开的论文“建立全景图像镶嵌及纹理映射模型(Creating full viewpanoramic image mosaics and texture-mapped models)”(ComputerGraphics(SIGRAPH′97)),第251~258页,1997年);陈(Chen)等公开的论文“Quicktime(Apple公司的数字多媒体技术)VR,虚拟环境导航的一种基于图像的方法(Quicktime VR an image-basedapproach to virtual environment navigation)”(Comp.Graph.Proc.,Annual Conf.Series(Siggraph′95),第29~38页,1995年);以及在1999年11月16日给予修切利斯基(Szeliski)等的题目为“用以构筑图像镶嵌的消重影方法及其设备(Deghosting method andapparatus for construction of image mosaics)”的美国专利第5,986,668号。 Refer as repair Rochelle Chomsky (Szeliski) and other public papers "Towards a panoramic image mosaic and texture mapped model (Creating full viewpanoramic image mosaics and texture-mapped models)" (ComputerGraphics (SIGRAPH'97)), 251 ~ 258 Page, 1997); Chen (Chen) and other published paper, "A Quicktime (Apple's digital media technology) VR, a virtual environment navigation method images (Quicktime VR an image-basedapproach to virtual environment navigation) based on the" ( Comp.Graph.Proc, Annual Conf.Series (Siggraph'95), 29 - 38, 1995);. and give repair Rochelle Chomsky (Szeliski), etc. in November 16, 1999, entitled "Use in order to build an image mosaic of deghosted method and apparatus (Deghosting method andapparatus for construction of image mosaics) "United States Patent No. 5,986,668.

由于提供了产生高分辨率、高亮度图像的实用解决手段,大型投影机阵列被大量采用。 By providing a practical means of settlement produce high resolution, high brightness, large projector arrays are widely adopted. 如在电视墙(video wall)的老式系统中典型的是用背面投影机的阵列,虽然邻接图像的边缘小,但通常存在可视见的间隙或重叠而被分离。 As typical in the TV wall (video wall) of the old system with an array of the back of the projector, while the edges of adjacent images of small, but usually see the presence of visible gaps or overlaps are separated. 而新型系统用叠加的投影机,为了进行图像的配准(registration)与混合(blending),设有通过手动操作进行准确的机电式调整的设备。 While the new system with superimposed projectors, in order to perform image registration (registration) and mixing (blending), it provided accurate adjustment of electromechanical devices by manual operation. 这些显示装置的设计仍旧非常单调,有必要使投影机准确地叠加,并且,多数情况下使其对显示表面大致垂直。 The design of these display devices is still very monotonous, it is necessary to accurately superimposed on the projector, and, in most cases it is substantially perpendicular to the display surface. 即,显示表面必须和各投影机的光轴垂直。 That is, the display surface must be perpendicular to the optical axis and each projector. 这肯定是大型显示装置设计中的最大缺点。 This is certainly a large display device design biggest drawback.

该图像投影过程可用已知的针孔相机模型来表现,在此点上投影机和照相机是类似的。 The image projection process can be aware of the pinhole camera model to show, at this point the projector and the camera are similar. 但是,过去的投影机在计算机视觉领域几乎不受人关注。 However, the last projector in the computer vision field almost unnoticed.

另外,拉斯克(Rasker)等公开的论文“采用基于相机配准的多重投影机显示(Multi-projector displays using camera-basedregistration)”(IEEE Visualization,第161~168页,1999年)中叙述了为建立多重投影机显示系统的各种方法。 In addition, the Lasker (Rasker) and other public papers, "based on the registration of multiple camera projector display (Multi-projector displays using camera-basedregistration)" (IEEE Visualization, 161 ~ 168, 1999) are described as the establishment of multi-projector display systems in various ways. 这些方法提供了用立体照相机在共有的坐标系中全面确定显示表面和各个投影机的内部与外部参数的、对无缝(seamless)显示问题的整体解决手段。 These methods provide a common perspective camera coordinate system fully determine the display surface and the internal and external parameters of each projector, seamless (seamless) means that the problem of the overall solution. 其结果是对整体显示环境的总括的叙述。 As a result, the overall umbrella of the display environment narrative. 但是,该方法虽然考虑了整体性解决手段,但实现该方法需要计算上的工作量和资源,具有其特有的复杂性。 However, this method, although considered a holistic means of settlement, but implementation of the method needs to be calculated on the amount of work and resources, with its unique complexity.

在陈(Chen)等公开的论文“利用一种UN校准的照相机的高分辨率多重投影机显示的自动配准(Automatic alignment of high-resolution multi-projector displays using an UN-calibrated camera)”(的文献(IEEE Visualization2000,2000年)中,提出了用可设置在镜头倾斜(pan-tilt)装置上的有可控的连续变焦与焦点的照相机,来减少机械的配准问题的机构。采用该机构,需要30分钟以上的时间进行数据的收集与计算。而在西亚替(Surety)的论文“用于无缝的大型显示的可缩放自调整显示技术(Scalable Self-Calibration DisplayTechnology for seamless large-scale Displays)”(pH论文,麻省理工学院的科技,1999年)中,也提供使用照相机来建立对多个投影机的相对位置的几何构造的解决方案。该照相机通过设于显示表面的正面的、一张印有规则的间隔配置的格子上成像进行校准。其后,将投影机图像在格子上进行配准。 Chen (Chen) and other published paper, "Automatic Registration UN calibrated using a high-resolution camera multiple projector display (Automatic alignment of high-resolution multi-projector displays using an UN-calibrated camera)" (the Document (IEEE Visualization2000, 2000), the proposed use can be set in the lens tilt (pan-tilt) have controlled continuous zoom and focus the camera on the device, to reduce the problem of registration agency machinery. With this mechanism , take 30 minutes or more to collect and calculate the data, while in the alternative (Surety) Valencia papers "for seamless large display of scalable self-adjusting display technology (Scalable Self-Calibration DisplayTechnology for seamless large-scale Displays ) "(pH papers, Massachusetts Institute of Technology, 1999), it is also provided with the camera to establish the relative positions of the plurality of the projector geometric configuration solution. The camera is provided on the display surface by positive, a printed image is calibrated on a regular lattice spacing configuration. Thereafter, the projector image registration is performed on the grid.

最好能明显地降低对投影机阵列的支持与成本。 The best can significantly reduce the projector arrays support cost. 并且,最好提供能适合给定投影机阵列结构的高自由度的图像配准与着色(rendering)技术。 And best offer to fit a high degree of freedom given image registration and coloring projector array structure (rendering) technology. 另外,最好完全自动化,并设置多个投影机,缩短为配准所需的时间。 In addition, the best fully automated, and set multiple projectors, to shorten the time required for the registration.

发明内容 DISCLOSURE

本发明提供一种设有任意个投影机的多重投影机显示系统。 The present invention provides a method has any number of projectors multiple projector display systems. 还提供一种对投影机阵列进行配准,用以形成一个无缝的长方形显示图像的自动化的方法。 The projector also provides an array of registration, to form a seamless rectangular automated method for image display. 用一台照相机,从投影的各图像的相对位置与方向判断投影机姿势。 With a camera, projector position is determined from the relative position and orientation of each image projected. 利用投影机的姿势,将投影机对平面的显示表面进行配准,达成图像的配准与浓度的混合。 Use position of the projector, the projector will be on display surface plane registration, reached the mixed image registration and concentration. 在工作中,在平面的显示表面上投影时,为了作为一幅图像被正确显示而源图像被翘曲。 In operation, when projected on the display surface plane, in order to be correctly displayed as an image and the source image is warped. 其后,该翘曲的图像被进行浓度修正以使其叠加相抵销。 Thereafter, the warped image density correction is performed so as to offset superposition phase. 然后,就能投影经修正的图像。 Then, the modified image can be projected. 本发明的优点之一是,显示表面可以相对投影机的光轴倾斜。 One of the advantages of the present invention is that the display surface can be tilted relative to the optical axis of the projector.

更具体地说,本发明是提供一种用多个投影机在显示表面上形成镶嵌图像的方法与系统。 More particularly, the present invention is a method and system for displaying a mosaic image formed on the surface to provide a plurality of projectors used. 各投影机依次在显示表面上投影配准图像,使被投影的配准图像结合成为多边形。 Each projector is in turn projected the image on the surface of registration, so that the projected image registration combined into polygons.

用照相机为各配准图像依次取得对应的输入图像。 With a camera image in order for the registration to obtain the corresponding input images. 然后,特别指定由多边形围成的显示表面的显示区域,并为各投影机确定显示区域和各输入图像之间的一个投影矩阵。 Then, specifying the display region surrounded by the polygon display surface, and a projection matrix region and between each of the input image is determined for the projector display.

用于各投影机的源图像,按照对应于投影机的一个投影矩阵翘曲。 For the source of each projector's image, according to the projector corresponds to a projection matrix warp. 该翘曲的源图像的像素随着投影矩阵叠加,然后将被翘曲、叠加的源图像同时向显示表面直接投影,形成镶嵌图像。 The warped source image pixels with the projection matrix stack, then the warped source image overlay, while direct projection to the display surface, forming a mosaic image.

按照本发明的一种用多个投影机在显示表面上形成镶嵌图像的方法,其特征在于包括:为所述各投影机依次在所述显示表面上投射配准图像,使该被投射的配准图像的组合体形成多边形的步骤;为所述各投射到所述显示表面的配准图像依次用照相机取得对应的输入图像的步骤;识别由所述多边形围成的所述显示表面上的显示区域的步骤;通过利用从所述输入图象中提取出来的单个的配准图像,在照相机和投影机之间为每个投影机定义一个第一单对应性的步骤;通过利用投射到显示区域以外并重叠到其他投射的图像的部分配准图像,在显示区域和照相机之间为每个投影机定义一个第二单对应性的步骤;通过将所述第一和第二单对应性组合在一起,为所述各投影机确定所述显示区域和所述各输入图像之间的单个的投影矩阵的步骤;为所述各投影机把源图像按照所述对应的一个投影矩阵翘曲的步骤;按照所述一个投影矩阵把所述翘曲的源图像的像素加权的步骤;以及直接在所述显示表面上同时投射所述翘曲的加权的源图像,并形成所述镶嵌图像的步骤。 According to the present invention, a method of using the mosaic image formed on the plurality of projectors display surface, comprising: sequentially to each of said projectors projecting said display image on the surface of registration, so that the projected feature Step polygon assembly of an image forming quasi; as projected onto the respective steps of the registration of the image display surface corresponding to the input image sequentially acquired by a camera; displaying on the surface of the polygon identified by the display surrounded By using the extracted from the input image out of a single registration images between the camera and projector in a first homography defined steps for each projector;; step by using the projected area of the display area outside and overlapped portion allocated to other registered images projected image between the display area and the camera of a second homography defined steps for each projector; by the first and second homography combination Together, each of the projector and the step of determining a single region of each of the input image between the display projection matrix; each of the source image to the projector according to the projection matrix corresponding to one warp step ; the projection matrix in accordance with said one of said step of warping the source image pixel weighting; and a step of weighting the source image while projecting the warped upper surface, and the mosaic image is formed directly on the display.

按照本发明的一种用以在显示表面上形成镶嵌图像的系统,其中包括:多个投影机,每个投影机用来将配准图像投射到所述显示表面上,使所述被投射的配准图像的组合体形成多边形;照相机,所述照相机用来为所述投射到所述显示表面的各配准图像依次取得对应的输入图像;第一定义单元,用来通过利用从所述输入图象中提取出来的单个的配准图像,在照相机和投影机之间为每个投影机定义一个第一单应性;第二定义单元,用来通过利用投射到显示区域以外并重叠到其他投射的图像的部分配准图像,在显示区域和照相机之间为每个投影机定义一个第二单对应性;识别单元,用以识别由所述多边形围成的所述显示表面上的显示区域;确定单元,用来为所述各投影机,在所述显示区域和所述各输入图像之间确定单独的投影矩阵;翘曲单元,用来为所述各投影机使源图像按照所述对应的单独的投影矩阵翘曲;以及加权装置,用来按照所述单独的投影矩阵,对翘曲的所述各源图像的像素加权,并把所述投影的、翘曲的和加权的源图像在所述显示表面上形成所述镶嵌图像。 According to the present invention provides a system for forming a mosaic image on the display surface, including: a plurality of projectors, each projector for the registration image is projected onto the display surface, said the projected The combination of the image forming polygon registration; camera, the camera used to display the projected onto the surface of each of the registration image corresponding to the input image acquired sequentially; a first defined cell, for use by the input from the extracted from a single image of the registration image, between the first camera and a projector homography is defined for each projector; second defined cell for use by projecting beyond the display area and overlapped to the other registration of images projected image distribution section, between the display area and a second camera homography is defined for each projector; identification unit for identifying a display area on the display surface of the polygon surrounded by the said ; determining a unit for each of the projectors in the display between the regions and the respective input image is determined solely projection matrix; warp unit, used to each projector so that the source of the image according to the separate projection matrix corresponding warp; and weighting means for projection matrix according to the individual, to the warped image pixel weighting of each source, and said projection, warped and weighted source The mosaic image is an image formed on the surface of the display.

附图说明 Brief Description

图1是本发明的多重投影机显示系统的示意图。 Figure 1 is a schematic view of multiple projector display system of the present invention.

图2是图1的显示系统的预处理阶段的流程图。 2 is a flowchart of the preprocessing stage display system 1 of FIG.

图3是图1的显示系统的着色阶段的流程图。 3 is a flowchart coloring stage display system of FIG. 1.

具体实施方式 DETAILED DESCRIPTION

定义下面的说明中,对后述的计算机操作的图像加以具体识别与定义。 Definitions The following description, to be specific recognition and definition of the images on the computer later operation.

配准图像(Registration image):投影在平面的显示表面的、有特定结构与几何形状的图像。 Image registration (Registration image): the projection plane of the display surface, an image of a specific structure and geometry. 多个配准图像被依次投影,投影机阵列内的每个投影机投影一个。 A plurality of registration images are projected sequentially, each projection projector a projection within the array.

相机图像(Camera image):在配准图像被投影期间,显示表面的由照相机拍摄的图像。 Camera image (Camera image): During the registration image is projected, the display image taken by the camera surface. 一个投影机投影的配准图像取一个相机图像。 A projector projected image registration to take a camera image.

源图像(Source image):为了形成镶嵌图像,应被投影在显示表面上的图像。 Source image (Source image): In order to form a mosaic image, the image should be projected on the display surface. 多个源图像同时产生,投影机的阵列内的投影机各一个。 A plurality of source images while producing projector projector arrays within each one.

被投影图像(Projected image):翘曲与混合后的源图像。 Projected image (Projected image): warp and mixed source image.

被显示图像(Displayed image):在显示表面上被投影的多个图像中由用户视觉感受的图像。 The displayed image (Displayed image): a plurality of images displayed on the surface of the image projected by the user visual experience.

显示区域(Display area):出现被显示图像的显示表面区域。 Display area (Display area): surface area appears to be a display image.

多重投影系统概述图1表示本发明的自修正型多重投影机显示系统100。 Multiple projection system overview Figure 1 shows a self-correction type multi-projector display system 100 of the present invention. 系统100包括投影机101~104的阵列和连接在处理器120的照相机110。 System 100 includes an array 101 to the projector 104 and the camera 120 is connected to the processor 110. 例如,本发明中用四台分辨率为1024768像素的三菱电机制X-80投影机和一台分辨率为640480像素的Logitech Quickcam Pro USB照相机。 For example, in this invention four pixel resolution of 1024 768 Mitsubishi Electric Corporation X-80 projector and a resolution of 640 480 pixels of Logitech Quickcam Pro USB camera. 系统100可用于前面或背面投影方式。 The system 100 may be used in front or rear projection.

投影机101~104可以为任意数量,任意地朝向平面形显示表面130设置,使投影的图像131~134大致重叠。 Projector 101 to 104 can be any number of arbitrary planar display surface 130 toward the settings to make the projected image roughly 131 to 134 overlapping. 要注意的是图像131~134不必一定配准或完全重叠。 It is noted that the image 131 to 134 are not necessarily registration or completely overlap. 实际上,投影的图像131~134产生梯形畸变,且互相旋转或倾斜,会在图像间生成间隙。 In fact, the projected image 131 ~ 134 generates trapezoidal distortion, or inclined to each other and rotating, a gap is generated between images. 而且,投影机的光轴可以对显示表面倾斜且互相间倾斜。 Moreover, the optical axis of the projector can display surface inclined to tilt with each other.

这里,投影机以22的阵列进行叠堆,但其它结构也可以。 Here, the projector to be 2 2 array stack, but other configurations may be. 例如,所有投影机可以在垂直方向叠堆或在水平方向排列。 For example, all projectors can stack or aligned horizontally in the vertical direction. 必需的只是,投影机要大致向着显示表面130,整体上显示区域135中有几个叠加的图像。 Required only, toward the display surface projection confidential roughly 130, 135 area there are several superimposed images on the display as a whole. 同样,照相机110的朝向被适当设置,以使其视野至少覆盖图像131~134的组合体。 Similarly, toward the camera 110 are appropriately set, so the image field of view to cover at least 131 to 134 of the assembly. 如下面更详细的叙述,要注意的是,照相机仅在预处理阶段200中使用。 As described in more detail below, note that the camera uses only 200 in the preprocessing stage. 在最佳实施例中,投影机与照相机均为数字装置。 In the preferred embodiment, the projector and the camera are digital devices.

处理器120中设有和用户相互交流的输入/输出(I/O)装置121。 Processor 120 features and user interaction of input / output (I / O) device 121. 它们是标准的结构。 They are standard structures. 并且,处理器还包含存于存储器内的可执行程序100,用以在工作中实施系统100的控制的方法。 And the processor is further stored in the memory comprising executable program 100, a method for controlling a system embodiment 100 at work. 使系统工作的方法包括预处理阶段200和着色阶段300。 Make the system work methods include pre-processing stage 200 and 300 colored stage. 这些在下面更详细描述。 These are described in more detail below. 如采用本发明的自修正,则在投影机阵列中,不管阵列的每个投影机以什么角度向平面形表面倾斜,都可以形成已知的形状,如一个以特定长宽比的长方形显示图像显示的镶嵌图像。 Such as the use of the present invention, since the correction in the projector in the array, each projector array regardless of what inclined at an angle to the planar surface, may be formed of a known shape, such as a display image in a particular aspect ratio rectangular displayed mosaic image.

配准与图像镶嵌方法本发明不是根据装置的类型和其相对的姿势两方面来确定其固有参数的投影机-照相机系统100来进行全部的配准,而是以多重投影机系统100,仅用一个投影矩阵,将来自各投影机的源图像翘曲。 Registration and image mosaic method of the present invention is not based on the type of device and the relative position of both to determine the intrinsic parameters of the projector - camera system 100 for all registration, but in multi-projector system 100, only a projection matrix from the source image warping each projector. 一个投影矩阵是按如下方式预先确定的、照相机和投影机之间的单对应性和显示区域和照相机之间的单对应性的组合。 A projection matrix is a combination of single homography between regions and the camera follows a pre-determined, cameras and projectors and show the correspondence between.

预处理图2表示预处理200中进行的过程。 Pretreatment Figure 2 shows a pretreatment process 200 performed. 配准图形,如棋盘图形140,为各投影机101~104而被依次投影在显示表面上(210)。 Registration pattern such as a checkerboard pattern 140, for the projector 101 to 104 are sequentially displayed on the projected surface (210). 配准图形的特征如角或线对应(C)221从输入图像111抽出。 Registration graphics features such as angles or lines correspond to (C) 221 is extracted from the input image 111. 本发明用48个对应达成子像素的配准精度。 48 correspond with the present invention to achieve sub-pixel registration accuracy. 对应211用于确定照相机和各投影机i之间的单对应性。 211 is used to determine the corresponding single camera and all correspondence between the projector i. (Hci)用于确定231。 (Hci) is used to determine 231.

知道所有因投影机的单对应性后,确定可利用显示区域135的尺寸(D)241(240)。 Know all correspondence because after a single projector, can be used to determine the size of the display area 135 (D) 241 (240). 尺寸241由被投影的所有图像131~134的组合体所围成的最大长方形(或者其它任意特定的形状)构成。 241 by the size of the projected image of all combinations of 131 ~ 134 surrounded by the largest rectangle (or any other special shapes) constituting. 要注意,显示区域会有其它规则或不规则的预先确定的形状,如圆形、三角形、椭圆形、菱形、脸面形(face)等。 Note that, the predetermined shape of the display area have other regular or irregular, such as round, triangular, oval, diamond, shaped face (face) and the like. 该形状可以由参数或形状标记划定。 The shape may be delineated by a parameter or shape mark.

知道可利用显示区域135的尺寸后,可指定在显示区域的外侧和其它被投影的图像叠加的被投影图像的部分。 Know that they can take advantage of the size of the display area 135, specifies the projected image is partially outside the display area of the projected image, and other overlay. 从而,显示坐标和照相机坐标之间的单对应性(Hrc)251和着色阶段中必需应用的浓度的权重(W)261(250)可以被确定。 Thus, the right to display the homography between the camera coordinates and the coordinates (Hrc) 251, and coloring stages required concentration of the applied weight (W) 261 (250) can be determined.

在显示区域外侧的像素有0的权重(黑),叠加部分的像素有0<W<1范围内的权重,其它全部的像素有1的权重。 Pixel in the display area outside has the right weight 0 (black), superimposing section pixel and 0 <W <right within a range of weight, all of the other pixels have a weight of 1. 由下面所述,在混合中各像素与其对应的权重相乘。 Described by the following, each pixel in the corresponding mixing weight multiplied.

着色如图3所示的着色处理(300)中,应被投影的源图像(I)301最初按照一个投影矩阵Bi302翘曲(310)。 FIG coloring coloring treatment (300) shown in Figure 3, the source image to be projected (I) 301 according to a first projection matrix Bi302 warpage (310). 然后,该被翘曲图像I311向I321混合(320)的过程中进行浓度加权,投影在显示区域135上(330)。 Then, the image is warped I321 I311 were mixed to a concentration of weighting (320) the process, the projection 135 on the display area (330). 浓度加权按像素各自进行。 Concentration were weighted according to each pixel. 例如,权重是在传统的着色处理装置中使用的α映像内存储。 For example, the weight is within the α used in conventional image processing device stores coloring.

单对应性为了定义着色中使用的一个投影矩阵Bi302,叙述了如何使用两个单对应性231及251。 A projection matrix Bi302 homography coloring used in order to define, describe how to use the two single correspondence between 231 and 251. 本发明采用针孔照相机模型。 The present invention uses a pinhole camera model. 两个照相机取得三维平面II内的两个图像时,两个图像由按比例(~=)定义的单对应性H而相关联。 When the camera takes two images in the two dimensional plane II, the correspondence between the two images from a single H rata (~ =) and associated definitions. 当m1和m2为平面II内的三维位置M的投影时,m2~=Hm1。 When m1 and m2 within the three-dimensional position of the plane II M projection, m2 ~ = Hm1. 式中,m1和m2表示齐次坐标,~=表示按比例相等。 Where, m1 and m2 represent homogeneous coordinates, ~ = represent proportionally the same.

本发明的多重投影机系统100中,最好使来自多个投影机Pi(式中,i=1,…,N)的N个源图像表示为无缝地接合的一个图像。 Multiple projector system 100 of the present invention, it is preferable that a plurality of N source images from the projectors Pi (where, i = 1, ..., N) is represented as an image seamlessly joined. 为了使投影机相互配准,本发明用一个照相机110取得全部N个各个配准图像的的输入图像。 In order to make the projector mutual registration, the present invention achieves all the N each registered images of the input image with a camera 110. 然后,投影机对照相机的映射与相对投影机之间的映射,用因使用平面显示表面130而产生的上述单对应性描述。 The projector mapping mapping between the projector relative to the camera, with the above description of the use of the corresponding single flat display surface 130 generated.

本发明,在表达为二维照相机坐标Xc=(x,y,1)Tes时,以及来自多个投影机的二维投影机坐标ui=(u,v,1)T时(i=1,…,N),均选择齐次坐标。 The present invention, in the expression for the two-dimensional camera coordinates Xc = (x, y, 1) Tes time, as well as two-dimensional coordinates of the projector from the plurality of projector ui = (u, v, 1) T when (i = 1, ..., N), have chosen to homogeneous coordinates. 在这样的场合,N个被投影的图像用满足下式的已知单对应性Hci,Hc2,…,HcN映射为输入图像131。 In such a case, N a projected image satisfies the following formula with known homography Hci, Hc2, ..., HcN 131 maps the input image.

ui~=HciXc(1)(式中,i=1,…,N)表示的图像中,被投影的图像131~134以已知的形状切割。 ui ~ = HciXc (1) (where, i = 1, ..., N) represents the image, the projected image 131 ~ 134 cut in a known shape. 和被投影的图像坐标等价的表示该形状内部的图像的坐标由Xr=(x,y,1)T表示。 And image coordinates are coordinates indicating equivalent projected image of the inside of the shape by Xr = (x, y, 1) T represents.

显示坐标和照相机坐标之间的关系可由第二二维投影矩阵Hrc描述。 It shows the relationship between the coordinates and camera coordinates by the second two-dimensional projection matrix Hrc description. 因此,下式成立。 Therefore, the following equation holds.

ui~=HciXc~=(HciHrc)Xr(2)(式中,i=1,…,N)为了简单表示,定义一组单对应性。 ui ~ = HciXc ~ = (HciHrc) Xr (2) (where, i = 1, ..., N) represents the sake of simplicity, the definition of a set of single correspondence.

Hri=HciHrc(式中,i=1,…,N)上式直接指定各投影机和显示的图像之间的几何关系。 Hri = HciHrc (where, i = 1, ..., N) on the formula specified geometry direct relationship between the projector and the image display between. 此时,两个任意投影的图像间的像素映射如下。 In this case, the image pixel mapping between two arbitrary projection follows.

uj~=HrjHri-1ui(3)式中,ui与uj各自表示投影机Pi与Pj的源图像内的对应像素。 uj ~ = HrjHri-1ui (3) where, ui and uj each represents a corresponding pixel of the source image projector Pi and Pj within.

配准为了形成无缝的显示图像,使被投影的图像相互配准,且使遍及叠加区域的浓度看起来柔和。 Registration In order to form a seamless display image, so the projected image registration each other, and so the concentration throughout the superimposed region looks soft. 本说明书中第一次叙述了将多个投影机101~104和一个照相机110用于配准的技术,然后,叙述了柔和的浓度混合(320)。 This specification describes the first plurality of projectors 101 to 104 and a camera 110 technology for registration, then, describes the gentle mixing concentration (320).

本小节中记载了为确定被投影图像间的关系的过程。 This section describes a process to determine the relationship between the projected image. 该关系是通过堆源图像301的翘曲处理,为了确定被投影的图像适当地配准,在着色(300)中被隐含地使用。 The relationship between the source image by warping the stack 301 process, in order to determine the projected image suitably registration, is used implicitly in coloration (300). 为被投影的镶嵌图像,本发明并不像传统技术那样为了无缝地接合被投影图像而去明显地校正照相机或投影机中的任一个;而是采用多个投影机和静止照相机之间的单对应性使多个被投影的图像充分正确配准的方法。 Mosaic image is being projected, the present invention is not like the traditional technology that seamlessly engaged to be projected away significantly corrected image camera or projector any one; but between the use of multiple projectors and still cameras single correspondence between the plurality of the projected image registration method fully configured correctly. 换言之,本发明并不需要像传统技术那样,为了确定内部与外部的参数而校正光学系统;本发明只依赖于各种单对应性。 In other words, the present invention does not require the conventional art as above, in order to determine the internal and external parameters correction optical system; the present invention only depend on various homography.

为了使投影机101~104和照相机110配准,本发明最初从各投影机101~104依次在显示表面130上投影配准图形140(210),并取得对应的输入图像111。 In order to make the projector 101 to 104 and the camera 110 registration, the present invention is originally from each projector 101 to 104 sequentially projected on the display surface 130 registration Graphics 140 (210), and acquires the corresponding input image 111. 通过从配准图形内已知的二维特征相对应的二维输入图像111抽出特征,本发明可基于式(1)确定照相机和各投影机之间的单对应性。 From registration by known 2D graphics features corresponding to the two-dimensional input image feature extraction 111, the present invention may be based on the formula (1) to determine the single camera and all correspondence between the projector.

更形式化地,将照相机110和给定的投影机之间的单对应性231,可作为最大似然估计问题而被公式化。 More formally, the single-camera 110 and 231 correspondence given between the projector can be used as maximum likelihood estimation problem is formulated. 输入图像x和配准图像u之间给出n个对应特征时,其最大似然值由以下的费用函数取最小值而求得。 Enter n when given a corresponding feature between the image and the registered image x u, the maximum likelihood value determined by the following cost function takes a minimum value.

&Sigma;i=1n||uj-HjXj||---(4)]]>式中,H是按比例地从照相机到投影机的单对应性。 & Sigma; i = 1n || uj-HjXj || --- (4)]]> where, H is proportionally from the camera to the projector homography. 相机图像和配准图像之间用4个以上对应时,H的8个未知参数由最小二乘法判断。 With four or more correspondence between the camera image and registration image, eight unknown parameters H judging by the least squares method. 可以用更复杂的技术改进对噪声或异常值的耐受性。 We can use more sophisticated techniques to improve noise or outliers tolerance. 可在任意场合,进行自动配准。 In any case, automated registration. 再有,如上所述,任意两个投影机间的相对的映射,可接合式(3)和(4)来确定。 Further, as described above, the relative mapping between any two projectors, and (4) to determine the engagement of the formula (3). 再请留意,这里不需要作明显的校正。 Then please note that there is no need for a significant correction.

为了形成被显示图像135,本发明是从所有投影机同时投影而形成的图像组合体所围成的最大尺寸的形状,例如可找出的长方形。 In order to form a display image 135, the present invention is a combination of the shape of the image is formed while projected from all projectors surrounded by the largest dimension, for example rectangular identified. 此问题可公式化为约束线性优化问题。 This problem can be formulated as a constrained linear optimization problems.

被照明的四边形131~134,在照相机110的坐标系中指定。 Illuminated quadrilateral 131 to 134, specified in the camera coordinate system 110. 本发明中,例如,最好能找到由N个四边形的组合体围成的给定长宽比构成的轴上配准的最大四边形。 The present invention, for example, it is best to find the maximum registration quadrilateral axis of N quadrilateral enclosed assembly composed of a given aspect ratio. 该组合体是在图1中由粗边表示的多边形L136。 The combination of a polygonal L136 in Fig. 1 represented by the rough edges. 本发明中,假设最大的长方形的至少一个顶点在多边形L的边上。 In the present invention, it is assumed the largest rectangular least one vertex in the polygon edge L. 由于长宽比固定,该长方形有三个自由度,即二维平面内的一个顶点的位置,以及长方形的比例。 Since the aspect ratio is fixed, the rectangle with three degrees of freedom, namely a two-dimensional plane of the vertex position, as well as rectangular proportions. 因此,本发明中,将多边形L136的边分离,通过检出在L边上存在长方形顶点的各可取位置和各可取的比例,求取最大的长方形。 Accordingly, the present invention, the separation of the polygon edges L136, desirable location and desirability of each vertex of each rectangle ratio by detecting the presence of L edge to strike the largest rectangle. 通过检查判断长方形的边和多边形L的边是否交叉。 By examining judge rectangular sides and edges of the polygon L is crossed. 发现大致内接的最合适的长方形后,用式(2)更新从显示坐标系到投影机像素的单对应性。 Found that the most suitable after the inscribed rectangle roughly by the formula (2) Update from the display coordinate system to a single projector pixel correspondence.

并且,即使一个照相机110不能视见所有被投影的图像,也可以通过添加更多的投影机,建立可伸缩的(scalable)的多重投影机系统100。 And even if a camera 110 can not be regarded see all the projected image, you can also add more projectors, create multiple projector system 100 scalable (scalable) a. 此时的必要条件,只是定义一个任意的投影机和与被显示图像135对应的坐标之间的关系。 Necessary condition at this time, but to define and coordinate the relationship between an arbitrary projector and 135 corresponding to the displayed image. 本最佳实施例中,以内接在视见的照相机内的长方形来定义该关系。 The preferred embodiment, less then a rectangle inside the view to see the camera to define the relationship. 但是,可以利用一个或多个照相机和显示坐标之间在欧几里得结构中的任意映射。 However, you can use one or more cameras and display any mapping structure between Euclidean coordinates. 因此,即使一个照相机不能视见所有的投影,只要有任意的一个照相机中视见两边的投影,就容易扩大邻接的投影机i和j之间的关系(HrjHri-1)。 Therefore, even if a camera can not see all the visual projection, as long as there is any one camera CTV see both sides of the projection, it is easy to expand relations adjoining projector i and j (HrjHri-1).

着色被显示的图像135由于各个投影机粗略设置而倾斜投影时,也作为多个被投影的图像321的按比例的无缝的图像(vision)被着色处理(300)。 135 colored image is displayed by setting individual projector rough tilt projection, also projected image as a plurality of 321 images by seamlessly proportion (vision) is colored handle (300). 如图3所示,本发明最初把按照投影矩阵302投影的各源图像301翘曲,投影在显示表面130上时,被翘曲的所有图像在显示区域135中看起来是配准的。 3, according to the present invention, the first projection matrix for each source image 301 projected warp 302, the projection 130 on the upper surface of the display, all the images are warped in the display area 135 seems to registration. 被翘曲的各图像311,之后用权重361进行混合处理,产生被投影的图像321,然后把该图像传送给投影机101~104中的一个。 Each image 311 is warped, followed by the right to be re-mixing process 361 to produce the projected image 321, and then the image is transmitted to the projector 101 to 104 in one.

利用单对应性的翘曲为投影图像所必需的翘曲处理,用将投影的图像的像素坐标变换为想要显示的图像的像素坐标的投影矩阵来加以定义。 Use homography warp of the projected image warping necessary, with the pixel coordinate conversion projected image of a projected image you want to display a matrix pixel coordinates to be defined. 如上所述,投影矩阵302在预处理中被确定。 As described above, the projection matrix 302 is determined during pretreatment. 在各投影机中,着色过程将各源图像301个别地加以翘曲。 In the projector, the coloring process will be warped image 301 each source individually.

源图像301被输入投影机的纹理存储器。 The source image 301 is input to the texture memory of the projector. 然后,在xy平面内,即具有范围[0∶1,0∶1]的单位长方形被纹理映射。 Then, in the xy plane, i.e., having a range of [0:1,0:1] The texture mapping unit is rectangular. 投影机i的场合,由于是xi~=Hriur,可用到以下的一个投影矩阵。 I projector occasions, because it is xi ~ = Hriur, available to one of the following projection matrix.

Bi=Hri[1|0] (5)式中,[1|0]用填充零矢量的单位矩阵将33矩阵变换为44矩阵。 Bi = Hri [1 | 0] (5) where, [1 | 0] matrix is filled with zero vector of the 3 3 matrix transform of 4 4 matrix. 请参照以下的式(6)。 Please refer to the following formula (6).

正确地说,本发明采用以Hri相乘的正射影矩阵。 Correct to say that the present invention uses a positive projection matrix multiplication with Hri. 显示坐标[0∶1,0∶1]的外侧区域内的像素,也出现在投影机帧缓冲器内。 Pixel display coordinates [0:1,0:1] outside the region, but also in the frame buffer inside the projector. 这虽然是在显示区域的外侧,但是和多边形L136即由N个照明的四边形的组合体所围成的区域相对应。 Although this is outside the display area, but that is by the polygon L136 combination of N lighting quadrilateral surrounded by the region, respectively.

背景色(0权重)被保持为黑色,以避免向该区域贡献。 The background color (0 weights) is held in black, in order to avoid the contribution to the region. 投影矩阵是为了计算深度缓冲,通过简单地增加行或列变换为在传统的图形流水线中使用的44矩阵。 The projection matrix is to calculate the depth buffer, by simply adding a row or column is converted to use in the traditional graphics pipeline 4 4 matrix. 结果的深度值当然可以不管,例如以下的公式成立。 The depth value of the result, of course, can not care, such as the establishment of the following formula.

[1|0]=1000010000100000---(6)]]>图像浓度的混合由多个投影机照明的显示区域135的区域,通常看起来突出地明亮。 [1 | 0] = 1000010000100000 --- region (6)]]> display area image density by mixing a plurality of lighting projectors 135, typically looks bright prominently. 为了使叠加的区域看起来无缝,即为了使具有柔和的浓度,要利用浓度混合过程。 In order to make the overlay area looks seamless, that in order to make a soft concentration, to take advantage of the concentration of the mixing process. 本发明中,为各投影机产生掩蔽(mask),通过该掩蔽给投影的图像内的所有像素分配范围为0.0到1.0的浓度权重。 The present invention, in order to produce each projector masking (mask), all pixels within the distribution range of the image projected by the mask to a concentration of 0.0 to 1.0 weight right.

混合处理中,将每个像素的浓度的权重装入到第二纹理映像的α通道上。 Mixing process, the concentration of the heavy weight of each pixel is loaded onto the α channel second texture image. 但是,色通道为黑色。 However, the black color channel. 相同的单位长方形由变更浓度权重的该纹理映像加以着色。 The same unit to be colored rectangle by the change in the concentration of weight texture mapping. α通道具有作为透过函数的作用,与翘曲的图像311的基底浓度相乘。 α channel has a role as a transmission function of substrate concentration and warped image 311 is multiplied. 如此,利用投影矩阵的翘曲与浓度修正这两项处理都可以用三维图形硬件有效率地完成。 Thus, warping and concentration correction by the projection matrix can both handle efficiently complete with three-dimensional graphics hardware.

请理解多个投影机在同一照明点上叠加的情况。 Please understand the situation multiple projectors on the same lighting point superimposed. 例如,在图1的显示区域135的中心,有四个投影机正好叠加的情况。 For example, the display area 135 in the center of FIG. 1, there is just the case of four projectors superimposed. 但是,为了得到均匀的浓度,使照明相同显示表面点的所有被投影的像素的权重之和成为1。 However, in order to obtain a uniform density, the same display illumination weights for all of the projected surface points of the pixel and the 1st.

简单的解决方法是仅用一个投影机照明给定的点。 The simplest solution is just a projector illumination given point. 另外的方法有,给予各投影机等量的权重,例如,两个投影机的场合分别给0.5的权重,四个投影机的场合分别给0.25的权重。 Another method is to give the right to the same amount of weight each projector, for example, the case of the two projectors were given weight of 0.5, four projectors occasions were 0.25 weight to right. 实际上由于投影机的配准、透视畸变以及非线性桶形畸变的小误差,被投影的图像在各边缘上严格说并不一致。 In fact due to the projector registration, perspective distortion and barrel distortion of the nonlinear error is small, the projected image on the edges, strictly speaking, is not consistent. 再有,经过一定时间投影机的位置会因机械或电气振动而改变。 Furthermore, after a certain period of time the position of the projector due to mechanical or electrical vibration change. 在存在上述任一问题时,即使一个图像的位置偏移也会在显示的图像内产生令人讨厌的线。 In the presence of any of the above problem, even if the position of an image shift line will produce objectionable in the displayed image. 因此,有必要在叠加中达到权重的柔和过渡。 Therefore, it is necessary to achieve a soft transition weights superposition. 结果,必须降低叠加的图像内的浓度受到静止配准与动态配准误差的影响。 As a result, the concentration must be reduced within the image superimposed still affected by registration and dynamic registration errors.

本发明的混合算法是采用边缘附近的像素上分配的权重大致为0的方式。 Hybrid algorithm of the present invention is to use a major cause right on the edge of the pixels near the allocation is zero way. 没有叠加的区域内的像素的权重显然为1,叠加区域内的邻接像素按如下方式分配其权重。 Right no overlapping pixels within the region is a heavy Obviously, adjacent pixel superimposition region follows its assigned weight.

更具体地说,本发明利用混合技术来发现和像素u=(u,v,1)相关联的权重Am(u)。 More particularly, the present invention uses hybrid technology to discover and pixel u = (u, v, 1) associated with the right weight Am (u). 于是,本发明对被翘曲的源图像311内的像素与到该图像边缘的距离成比例地进行加权。 Thus, the present invention is the distance of the pixel to the edge of the image to be warped source images 311 are weighted in proportion. 更严格地说,就是像素以与最近的“不可见”像素即权重为0的像素的距离成比例地进行加权。 More precisely, to the nearest pixel it is "invisible" to the weight of the pixel that is weighted in proportion to the distance from 0 pixels. 投影矩阵用归一化的投影机坐标进行计算,以使u与v坐标在[0,1]区间变化。 Projection matrix calculated using normalized coordinates projector, so that u and v coordinates [0,1] change. 因此,某一像素到投影机Pi内的最近边缘的距离由下式表述。 Therefore, a certain pixel from the nearest edge of the projector inside the Pi is expressed by the following formula.

di(u)=w(u,v)min(u,v,1-u,1-v) (7)式中,u∈[0,1]时w(u,v)=1,v∈[0,1]时w(u,v)=0。 di (u) = w (u, v) min (u, v, 1-u, 1-v) (7) wherein, u∈ [0,1] When w (u, v) = 1, v∈ [0,1] is w (u, v) = 0.

这就把权重分配问题简化为求最小函数。 This is the weight distribution problem is reduced to the minimum required functions. 再有,基于在多个投影机范围的隐含的像素对应,本发明使照明同一显示表面的像素的权重相加之和确实为一。 Further, the pixels corresponding to the implied multiple projectors based on the scope of the present invention, the illumination of the same weight of the sum of the display pixel and the right surface really is one. 与投影机Pm的像素u相关联的像素的权重Am(u)由以下式可计算。 Pm projector pixels associated with pixels u weight Am (u) can be calculated by the following formula.

Am(u)=dm(u)/(∑idi(HriHrm-1u)) (8)(式中,j=1,…,N)该类型的浓度混合处理,不管被投影的图像间的间隙或多余的叠加中产生不产生误差,存在较小的位置偏移误差时其稳定性变高。 Am (u) = dm (u) / (Σidi (HriHrm-1u)) (8) (where, j = 1, ..., N) concentration of the mixed handle the type, or whether the gap between the projected image superposition excess produce errors, the presence of a small offset error position stability becomes higher. 如下所述,在三维图形硬件中为得到透明度,用共同使用的α映射导入权重。 As described below, in order to obtain three-dimensional graphics hardware transparency, together with the use of mapping α import weights. 最后,被翘曲且浓度被加权的图像321由各投影机投影(330),产生显示图像。 Finally, the warp and the concentration is weighted by the projector projection image 321 (330), resulting in the display image.

[发明效果]对于在短时间内设定多个数字投影机,使组合的投影在图像内不发生明显的重叠且显示为一个组合图像,本发明是有效的。 [Effects] is set for more than a short time in a digital projector so that the combined image is not projected in the occurrence of significant overlap and appear as a combination of image, the present invention is effective. 投影机和一个处理器连接。 And a processor connected to the projector. 用户只是简单地使投影机全部朝向显示表面的方向,使各个被投影的图像以某一预定的方式,如有六台投影机的时候以23的矩阵叠加或大致叠加。 Users simply allows the projector to display the direction toward the surface of all, so that each projected image with a predetermined manner, if six projectors when 2 3 matrix superposition or substantially superimposed. 一旦投影机被粗略地设置,由于使投影机彼此间且投影机和显示表面等配准,可由各投影机显示配准图像。 Once the projector is set roughly, since the projector and the projector to each other and the display surface registration can be displayed by the projector image registration. 如投影机被配准,就立即可用该投影机显示无缝的镶嵌图像。 As the projector is registration, it is immediately available to the projector to display a seamless mosaic image. 整个过程只需2、3秒。 The whole process takes 2,3 seconds.

本发明用特定的用语与示例叙述。 The present invention is described using specific terms and examples. 本发明的精神与范围内,要知道还会有其它各种适合的方案与变更的方案。 Within the spirit and scope of the invention, to know there will be a program of appropriate programs and various other changes. 因此,附加权利要求书的目的在于,覆盖所有符合本发明的精神和在本发明范围内的所有变形与变更。 Therefore, the appended claims aims to cover all eligible spirit of the invention and all such modifications and changes within the scope of the invention.

Classifications
International ClassificationH04N17/00, H04N9/12, H04N5/74
Cooperative ClassificationH04N9/12, H04N9/3147, H04N9/3194
European ClassificationH04N9/31R3, H04N9/31T1, H04N9/12
Legal Events
DateCodeEventDescription
31 Dec 2003C06Publication
10 Mar 2004C10Entry into substantive examination
1 Jun 2005C14Grant of patent or utility model
30 Sep 2015EXPYTermination of patent right or utility model