WO2008147039A1 - System and method for recognizing images using t-test - Google Patents

System and method for recognizing images using t-test Download PDF

Info

Publication number
WO2008147039A1
WO2008147039A1 PCT/KR2008/001665 KR2008001665W WO2008147039A1 WO 2008147039 A1 WO2008147039 A1 WO 2008147039A1 KR 2008001665 W KR2008001665 W KR 2008001665W WO 2008147039 A1 WO2008147039 A1 WO 2008147039A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
situations
module
recognizers
face images
Prior art date
Application number
PCT/KR2008/001665
Other languages
French (fr)
Inventor
Phill Kyu Rhee
Mi Young Nam
Original Assignee
Inha-Industry Partnership Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Inha-Industry Partnership Institute filed Critical Inha-Industry Partnership Institute
Publication of WO2008147039A1 publication Critical patent/WO2008147039A1/en

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/809—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of classification results, e.g. where the classifiers operate on the same input data

Definitions

  • the present invention relates to a system for recognizing images, and more particularly to a system for recognizing images using t-tests, which classifies recognizers using t-tests based on recognized image situations, so that recognizers capable of obtaining optimum image situations suitable for individual image situations can be selected, and face images having different image situations can be efficiently recognized through the classification of the image situations of the face images .
  • Pattern recognition refers to a process of performing mapping to class-membership space. That is, it is a process of extracting important features from data acquired from the outside, classifying the closest templates based on the extracted features, and obtaining a final result.
  • the main trend of the research is increasing the recognition rate by fusing multiple recognizers into a single system and combining output results together.
  • the research does not present solutions on to how to fuse the results into a final result, but is limited to which recognizers are suitable for use for individual data regions . Accordingly, research into methods of efficiently classifying data and methods of classifying data so that classified data has optimum similarities is not sufficient .
  • the accuracy of recognition can be guaranteed only for data regions classified under uniform conditions because the situation in which the individual data regions were classified is not considered, so that the range of application of the research is limited.
  • an object of the present invention is to provide a system for recognizing images using t-tests, which can classify recognizers using t-tests based on image situations recognized through the image situation recognition module, so that recognizers capable of obtaining optimal image situations suitable for individual image situations can be selected, and so that face images having different image situations can be efficiently recognized through the classification of the image situations of the face images .
  • the system including an image recognition module for loading face images,- an image normalization module for normalizing the loaded face images; an image pre-processing module for removing noise from the normalized face images and generating vector data,- an image situation classification module for classifying image situations, included in the vector data, into a user-specified number of types through clustering using a K-average clustering method; an image situation recognition module for calculating the average values of the classified image situations based on the image situations and center points using a Euclidean distance formula, and recognizing the image situations based on the calculated average values; and a recognizer fusion module for selecting recognizers suitable for the individual image situations through classification of the recognizers using t- tests based on the recognized image situations .
  • the present invention provides a method of recognizing face images using t-tests in an image classification method using recognizers, including a first step for loading face images; a second step for normalizing the loaded face images,- a third step for removing noise from the normalized face images, and generating vector data; a fourth step for classifying image situations, included in the vector data, into a user-specified number of types through clustering using a K-average clustering method; a fifth step for calculating the average values of the classified image situations based on the image situations and center points using a Euclidean distance formula, and recognizing image situations based on the calculated average values; and a sixth step for selecting recognizers suitable for the individual image situations through classification of the recognizers using t-tests based on the recognized image situations.
  • FIG. 1 is a diagram showing the structure of a conventional single recognizer
  • FIG. 2 is a flow diagram showing the fusion of multiple recognizers in which single recognizers are fused together;
  • FIG. 3 is a diagram showing the construction of a system for recognizing images using t-tests according to an embodiment of the present invention
  • FIG. 4 is a diagram showing the relationships among clusters according to an embodiment of the present invention.
  • FIG. 5 is a diagram showing the similarity tables of individual clusters according to an embodiment of the present invention
  • FIG. 6 is a diagram showing the classification of image situations through an image situation classification module according to an embodiment of the present invention
  • FIG. 7 is a diagram showing the recognition rates of individual recognizers according to an embodiment of the present invention.
  • FIG. 8 is a graph showing the results of the combination of two recognizers CLS A and CLS B according to an embodiment of the present invention.
  • FIG. 9 is graphs showing the results of (P A -S A ) , (P B ⁇ S B ) , and (P F -S F ) , obtained through Equation 4 according to an embodiment of the present invention.
  • FIG. 10 is a flowchart showing a method of recognizing images using t-tests according to an embodiment of the present invention.
  • an image recognition system 100 using t-tests includes an image recognition module 110, an image normalization module 120, an image pre-processing module 130, an image situation classification module 140, and a recognizer fusion module 160.
  • the image recognition system 100 loads face images of a predetermined size from an image capture device such as a digital camcorder.
  • the image normalization module 120 normalizes the loaded face images.
  • the "normalization” is set to the normalization of loaded images into images of a 128 x 128 size, but the present invention is not limited thereto.
  • the image pre-processing module 130 removes noise from the normalized images received from the image normalization module 120 through histogram equalization, and then generates vector data.
  • the image situation classification module 140 receives the noise-free vector data from the image pre-processing module 130, and classifies image situations into a user-specified number of types through clustering using a K-average clustering method.
  • the "image situations" have information that enables individual images to be distinguished from each other based on differences in the brightness (light and shade) of the individual images, and may be understood to be the illumination situations of the images that are obtained by setting the individual images to brightness (light and shade) values ranging from 0 to 255. The reason for this is to prevent a reduction in the performance of the recognition of face images attributable to the difference between image situations included in the face images .
  • clustering calculates similarities, output from individual classifiers, as the probabilities of matching the right person using the following Equation 1 and then determines whether the results output from individual classifiers are true or false based on the calculation results .
  • Image situations are classified through clustering, as shown in the attached FIG. 6.
  • the image situation recognition module 150 calculates average values based on image situations and center points using the Euclidean distance formula based on the image situations classified through the image situation classification module 140, and recognizes the image situations based on the calculated average values .
  • the recognizer fusion module 160 selects recognizers suitable for the image situations recognized through the classification of recognizers using t-tests based on the image situations recognized through the image situation recognition module 150. This selection of recognizers is performed through t-tests, and the selection of recognizers using the t- tests is performed as follows . First, it is assumed that there exist two image recognizers CLSA and CLSB. CLSA is defined as a distribution
  • A and a face situation distribution is defined as
  • F A ⁇ N(F A ,o A ) _ CLSB is defined as an image situation
  • FIG. 8 shows the results of the combination of the two recognizers, and the results are obtained through the following Equation 3 :
  • FIG. 9 shows the results of recognition rates (PA - SA) , (PB - SB) , and (PF - SF) of individual recognizers, which are obtained through the following Equation 4. It can be seen that the recognition rate (PF - SF) of a recognizer CLSF, into which the two recognizers CLSA and CLSB are combined, is higher than the best recognition rate (PA - SA) of the recognizer CLSA, as well • as the second best recognition rate (PB - SB) of the recognizer CLSB .
  • the distance between the recognizer CLSA, exhibiting the best recognition rate (PA - SA) , and the recognizer CLSB, exhibiting the second best recognition rate (PB - SB) is obtained using the following Equation 5 , and the standard deviation of the distance is defined as ⁇ ⁇ 0 . 6SA.
  • the recognizer fusion module 160 selects a recognizer CLSF into which the two recognizers CLSA and CLSB are combined based on the results of calculation using the following Equation 6 :
  • Recognizers that are suitable for the face situations recognized through the image situation recognition module 150 can be selected based on recognition rates that are obtained by the individual recognizers through the classification of the recognizers using t-tests, as described above.
  • the image recognition module 110 loads face images from an image-capture device at step SIlO, and the image normalization module 120 normalizes the loaded face images at step S120.
  • the image pre-processing module 130 removes noise from the normalized face images through histogram equalization, and generates vector data at step S130.
  • the image situation classification module 140 classifies image situations into a user-specified number of types through clustering using a K-average clustering method based on the noise-free vector data at step S140.
  • the image situations recognition module 150 calculates average values of the classified image situations based on the image situations and center points using the Euclidean distance formula, and recognizes the image situations based on the calculated average values at step
  • the recognizer fusion module 160 selects recognizers suitable for individual image situations through the classification of recognizers using t-tests based on the recognized image situations at step S160.
  • recognizers can be classified using t-tests based on image situations recognized through the image situation recognition module, so that recognizers capable of obtaining optimal image situations suitable for the individual image situations can be selected, and so that face images having different image situations can be efficiently recognized through the classification of the image situations of the face images .

Abstract

Disclosed herein is a system and method for recognizing images using t-tests. The system includes an image recognition module, an image normalization module, an image pre-processing module, an image situation classification module, an image situation recognition module, and a recognizer fusion module. The image recognition module loads face images. The image normalization module normalizes the loaded face images. The image pre-processing module removes noise from the normalized face images and generates vector data. The image situation classification module classifies image situations into a user-specified number of types through clustering using a K-average clustering method. The image situation recognition module calculates the average values of the classified image situations based on the image situations and center points using the Euclidean distance formula, and recognizes the image situations. The recognizer fusion module selects recognizers suitable for the individual image situations through classification of the recognizers using t-tests based on the recognized image situations.

Description

SYSTEM AMD METHOD FOR RECOGNIZING IMAGES USING T-TEST
Technical Field
The present invention relates to a system for recognizing images, and more particularly to a system for recognizing images using t-tests, which classifies recognizers using t-tests based on recognized image situations, so that recognizers capable of obtaining optimum image situations suitable for individual image situations can be selected, and face images having different image situations can be efficiently recognized through the classification of the image situations of the face images .
Background Art
Pattern recognition refers to a process of performing mapping to class-membership space. That is, it is a process of extracting important features from data acquired from the outside, classifying the closest templates based on the extracted features, and obtaining a final result.
Accordingly, it is very important in such pattern recognition to collect data suitable for a desired application field and select recognizers for performing recognition in consideration of the characteristics of the collected individual data. Basically, most recognition methods using recognizers employ a single recognizer structure, in which input data λx' is classified based on an output xy' through a single recognizer, as shown in FIG. 1.
However, since such a single recognizer has a problem in that the output result thereof varies with the type of data or the environment, a lot of research related to a multiple recognizer system capable of obtaining desired recognition results has been carried out recently.
However, as shown in the attached FIG. 2, the main trend of the research is increasing the recognition rate by fusing multiple recognizers into a single system and combining output results together. In the case in which a plurality of recognizers is selectively used, the research does not present solutions on to how to fuse the results into a final result, but is limited to which recognizers are suitable for use for individual data regions . Accordingly, research into methods of efficiently classifying data and methods of classifying data so that classified data has optimum similarities is not sufficient .
Furthermore, the accuracy of recognition can be guaranteed only for data regions classified under uniform conditions because the situation in which the individual data regions were classified is not considered, so that the range of application of the research is limited.
Disclosure of the Invention
Accordingly, the present invention has been made keeping in mind the above problems occurring in the prior art, and an object of the present invention is to provide a system for recognizing images using t-tests, which can classify recognizers using t-tests based on image situations recognized through the image situation recognition module, so that recognizers capable of obtaining optimal image situations suitable for individual image situations can be selected, and so that face images having different image situations can be efficiently recognized through the classification of the image situations of the face images .
In order to a system for recognizing images using t- tests, the system including an image recognition module for loading face images,- an image normalization module for normalizing the loaded face images; an image pre-processing module for removing noise from the normalized face images and generating vector data,- an image situation classification module for classifying image situations, included in the vector data, into a user-specified number of types through clustering using a K-average clustering method; an image situation recognition module for calculating the average values of the classified image situations based on the image situations and center points using a Euclidean distance formula, and recognizing the image situations based on the calculated average values; and a recognizer fusion module for selecting recognizers suitable for the individual image situations through classification of the recognizers using t- tests based on the recognized image situations .
Additionally, the present invention provides a method of recognizing face images using t-tests in an image classification method using recognizers, including a first step for loading face images; a second step for normalizing the loaded face images,- a third step for removing noise from the normalized face images, and generating vector data; a fourth step for classifying image situations, included in the vector data, into a user-specified number of types through clustering using a K-average clustering method; a fifth step for calculating the average values of the classified image situations based on the image situations and center points using a Euclidean distance formula, and recognizing image situations based on the calculated average values; and a sixth step for selecting recognizers suitable for the individual image situations through classification of the recognizers using t-tests based on the recognized image situations.
Brief Description of the Drawings
The above and other objects, features and other advantages of the present invention will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a diagram showing the structure of a conventional single recognizer;
FIG. 2 is a flow diagram showing the fusion of multiple recognizers in which single recognizers are fused together;
FIG. 3 is a diagram showing the construction of a system for recognizing images using t-tests according to an embodiment of the present invention;
FIG. 4 is a diagram showing the relationships among clusters according to an embodiment of the present invention;
FIG. 5 is a diagram showing the similarity tables of individual clusters according to an embodiment of the present invention; FIG. 6 is a diagram showing the classification of image situations through an image situation classification module according to an embodiment of the present invention;
FIG. 7 is a diagram showing the recognition rates of individual recognizers according to an embodiment of the present invention;
FIG. 8 is a graph showing the results of the combination of two recognizers CLSA and CLSB according to an embodiment of the present invention;
FIG. 9 is graphs showing the results of (PA-SA) , (PB~SB) , and (PF-SF) , obtained through Equation 4 according to an embodiment of the present invention; and
FIG. 10 is a flowchart showing a method of recognizing images using t-tests according to an embodiment of the present invention.
Best Mode for Carrying Out the Invention
The features and advantages of the present invention will become more apparent from the following detailed description with reference to the attached drawings. It is to be noted that, if it is determined that detailed descriptions of well-known functions related to the present invention make the gist of the present invention obscure, the detailed descriptions will be omitted below.
Referring to FIG. 3, an image recognition system 100 using t-tests according to the present invention includes an image recognition module 110, an image normalization module 120, an image pre-processing module 130, an image situation classification module 140, and a recognizer fusion module 160.
In detail, the image recognition system 100 loads face images of a predetermined size from an image capture device such as a digital camcorder. The image normalization module 120 normalizes the loaded face images. Here, the "normalization" is set to the normalization of loaded images into images of a 128 x 128 size, but the present invention is not limited thereto.
The image pre-processing module 130 removes noise from the normalized images received from the image normalization module 120 through histogram equalization, and then generates vector data.
Referring to the attached FIG. 4, the image situation classification module 140 receives the noise-free vector data from the image pre-processing module 130, and classifies image situations into a user-specified number of types through clustering using a K-average clustering method. Here, the "image situations" have information that enables individual images to be distinguished from each other based on differences in the brightness (light and shade) of the individual images, and may be understood to be the illumination situations of the images that are obtained by setting the individual images to brightness (light and shade) values ranging from 0 to 255. The reason for this is to prevent a reduction in the performance of the recognition of face images attributable to the difference between image situations included in the face images . Referring to the attached FIG. 5, clustering calculates similarities, output from individual classifiers, as the probabilities of matching the right person using the following Equation 1 and then determines whether the results output from individual classifiers are true or false based on the calculation results . Image situations are classified through clustering, as shown in the attached FIG. 6.
D(X)= {ω'm ,dm(*))o dM(x)= max(dj(*)) (1)
The image situation recognition module 150 calculates average values based on image situations and center points using the Euclidean distance formula based on the image situations classified through the image situation classification module 140, and recognizes the image situations based on the calculated average values .
As shown in the attached FIG. 7, since an image recognizer performs differently depending on the type of data used, the results of the recognition by individual recognizers are different, so that it can be seen that the best performance can be obtained using a method of averaging post- probabilities output from individual classifiers . The recognizer fusion module 160 selects recognizers suitable for the image situations recognized through the classification of recognizers using t-tests based on the image situations recognized through the image situation recognition module 150. This selection of recognizers is performed through t-tests, and the selection of recognizers using the t- tests is performed as follows . First, it is assumed that there exist two image recognizers CLSA and CLSB. CLSA is defined as a distribution
Λ p having an average of recognition rates A and a variance value
A , and a face situation distribution is defined as
Λ
FA~N(FA,oA) _ CLSB is defined as an image situation
Λ distribution PB ~ N(PB><>B) _ CLSA and CLSB are combined together, and the standard deviation between recognition rates for the two recognizers is obtained using the following Equation 2 :
δF=ja2δ2+b2δ2 (2)
The attached FIG. 8 shows the results of the combination of the two recognizers, and the results are obtained through the following Equation 3 :
Sp=^a2S2+b2SB 2 (3)
The attached FIG. 9 shows the results of recognition rates (PA - SA) , (PB - SB) , and (PF - SF) of individual recognizers, which are obtained through the following Equation 4. It can be seen that the recognition rate (PF - SF) of a recognizer CLSF, into which the two recognizers CLSA and CLSB are combined, is higher than the best recognition rate (PA - SA) of the recognizer CLSA, as well • as the second best recognition rate (PB - SB) of the recognizer CLSB .
aPA + bPB-^a2SA 2 +b2SB 2 > PA-S,
PF-SF > PA-SA
PF = aPA+bPB (4 )
The distance between the recognizer CLSA, exhibiting the best recognition rate (PA - SA) , and the recognizer CLSB, exhibiting the second best recognition rate (PB - SB) , is obtained using the following Equation 5 , and the standard deviation of the distance is defined as Δ < 0 . 6SA.
Λ Λ
Δ = PΛ-PB
Figure imgf000010_0001
The recognizer fusion module 160 selects a recognizer CLSF into which the two recognizers CLSA and CLSB are combined based on the results of calculation using the following Equation 6 :
°Λ ~~ '(0.05,W-I) N
Figure imgf000010_0002
Here, t (0.05,N-I) is the case in which t = 0.95, and t(0.05,N-l) = 1.96 is used if N > 100.
Recognizers that are suitable for the face situations recognized through the image situation recognition module 150 can be selected based on recognition rates that are obtained by the individual recognizers through the classification of the recognizers using t-tests, as described above.
Next, a method of recognizing images using t-tests according to an embodiment of the present invention will be described below. Referring to FIG. 10, the image recognition module 110 loads face images from an image-capture device at step SIlO, and the image normalization module 120 normalizes the loaded face images at step S120. The image pre-processing module 130 removes noise from the normalized face images through histogram equalization, and generates vector data at step S130.
The image situation classification module 140 classifies image situations into a user-specified number of types through clustering using a K-average clustering method based on the noise-free vector data at step S140.
Thereafter, the image situations recognition module 150 calculates average values of the classified image situations based on the image situations and center points using the Euclidean distance formula, and recognizes the image situations based on the calculated average values at step
S150.
The recognizer fusion module 160 selects recognizers suitable for individual image situations through the classification of recognizers using t-tests based on the recognized image situations at step S160.
Industrial Applicability According to the above-described present invention, recognizers can be classified using t-tests based on image situations recognized through the image situation recognition module, so that recognizers capable of obtaining optimal image situations suitable for the individual image situations can be selected, and so that face images having different image situations can be efficiently recognized through the classification of the image situations of the face images .
Although the preferred embodiments of the present invention have been disclosed for illustrative purposes, those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims .

Claims

Claims
1. A system for recognizing images using t-tests, the system comprising: an image recognition module for loading face images; an image normalization module for normalizing the loaded face images,- an image pre-processing module for removing noise from the normalized face images and generating vector data; an image situation classification module for classifying image situations, included in the vector data, into a user- specified number of types through clustering using a K-average clustering method; an image situation recognition module for calculating average values of the classified image situations based on the image situations and center points using a Euclidean distance formula, and recognizing the image situations based on the calculated average values; and a recognizer fusion module for selecting recognizers suitable for the individual image situations through classification of the recognizers using t-tests based on the recognized image situations .
2. The system as set forth in claim 1, wherein the image situations are obtained by setting the individual images to brightness values ranging from 0 to 255.
3. A method of recognizing face images using t-tests in an image classification method using recognizers, comprising: a first step for loading face images; a second step for normalizing the loaded face images; a third step for removing noise from the normalized face images, and generating vector data; a fourth step for classifying image situations, included in the vector data, into a user-specified number of types through clustering using a K-average clustering method; a fifth step for calculating average values of the classified image situations based on the image situations and center points using a Euclidean distance formula, and recognizing image situations based on the calculated average values ; and a sixth step for selecting recognizers suitable for the individual image situations through classification of the recognizers using t-tests based on the recognized image situations .
PCT/KR2008/001665 2007-05-25 2008-03-25 System and method for recognizing images using t-test WO2008147039A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR10-2007-0050779 2007-05-25
KR1020070050779A KR100870724B1 (en) 2007-05-25 2007-05-25 System for detecting image using t-test and method therefor

Publications (1)

Publication Number Publication Date
WO2008147039A1 true WO2008147039A1 (en) 2008-12-04

Family

ID=40075235

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2008/001665 WO2008147039A1 (en) 2007-05-25 2008-03-25 System and method for recognizing images using t-test

Country Status (2)

Country Link
KR (1) KR100870724B1 (en)
WO (1) WO2008147039A1 (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101984576A (en) * 2010-10-22 2011-03-09 北京工业大学 Method and system for authenticating anonymous identity based on face encryption
CN102156871A (en) * 2010-02-12 2011-08-17 中国科学院自动化研究所 Image classification method based on category correlated codebook and classifier voting strategy
ES2432479R1 (en) * 2012-06-01 2014-05-28 Universidad De Las Palmas De Gran Canaria Method for the identification and automatic classification of arachnid species through their spider webs
CN104036254A (en) * 2014-06-20 2014-09-10 成都凯智科技有限公司 Face recognition method
DE102015200434A1 (en) * 2015-01-14 2016-07-14 Bayerische Motoren Werke Aktiengesellschaft Method and apparatus for improving object recognition in different lighting situations
DE102015200433A1 (en) * 2015-01-14 2016-07-14 Bayerische Motoren Werke Aktiengesellschaft Method and apparatus for reducing the testing effort in the evaluation of an object recognition system
DE102015200437A1 (en) * 2015-01-14 2016-07-14 Bayerische Motoren Werke Aktiengesellschaft Method and device for determining the confidence of an object recognition
CN109977803A (en) * 2019-03-07 2019-07-05 北京超维度计算科技有限公司 A kind of face identification method based on Kmeans supervised learning

Families Citing this family (43)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9998697B2 (en) 2009-03-02 2018-06-12 Flir Systems, Inc. Systems and methods for monitoring vehicle occupants
US9473681B2 (en) 2011-06-10 2016-10-18 Flir Systems, Inc. Infrared camera system housing with metalized surface
US9843742B2 (en) 2009-03-02 2017-12-12 Flir Systems, Inc. Thermal image frame capture using de-aligned sensor array
US9948872B2 (en) 2009-03-02 2018-04-17 Flir Systems, Inc. Monitor and control systems and methods for occupant safety and energy efficiency of structures
US9756264B2 (en) 2009-03-02 2017-09-05 Flir Systems, Inc. Anomalous pixel detection
US9635285B2 (en) 2009-03-02 2017-04-25 Flir Systems, Inc. Infrared imaging enhancement with fusion
US9986175B2 (en) 2009-03-02 2018-05-29 Flir Systems, Inc. Device attachment with infrared imaging sensor
US9208542B2 (en) 2009-03-02 2015-12-08 Flir Systems, Inc. Pixel-wise noise reduction in thermal images
US9674458B2 (en) 2009-06-03 2017-06-06 Flir Systems, Inc. Smart surveillance camera systems and methods
US10757308B2 (en) 2009-03-02 2020-08-25 Flir Systems, Inc. Techniques for device attachment with dual band imaging sensor
WO2012170949A2 (en) 2011-06-10 2012-12-13 Flir Systems, Inc. Non-uniformity correction techniques for infrared imaging devices
US9451183B2 (en) 2009-03-02 2016-09-20 Flir Systems, Inc. Time spaced infrared image enhancement
US10244190B2 (en) 2009-03-02 2019-03-26 Flir Systems, Inc. Compact multi-spectrum imaging with fusion
US9235876B2 (en) 2009-03-02 2016-01-12 Flir Systems, Inc. Row and column noise reduction in thermal images
US9517679B2 (en) 2009-03-02 2016-12-13 Flir Systems, Inc. Systems and methods for monitoring vehicle occupants
USD765081S1 (en) 2012-05-25 2016-08-30 Flir Systems, Inc. Mobile communications device attachment with camera
US9292909B2 (en) 2009-06-03 2016-03-22 Flir Systems, Inc. Selective image correction for infrared imaging devices
US9716843B2 (en) 2009-06-03 2017-07-25 Flir Systems, Inc. Measurement device for electrical installations and related methods
US9756262B2 (en) 2009-06-03 2017-09-05 Flir Systems, Inc. Systems and methods for monitoring power systems
US9819880B2 (en) 2009-06-03 2017-11-14 Flir Systems, Inc. Systems and methods of suppressing sky regions in images
US10091439B2 (en) 2009-06-03 2018-10-02 Flir Systems, Inc. Imager with array of multiple infrared imaging modules
US9843743B2 (en) 2009-06-03 2017-12-12 Flir Systems, Inc. Infant monitoring systems and methods using thermal imaging
US9706138B2 (en) 2010-04-23 2017-07-11 Flir Systems, Inc. Hybrid infrared sensor array having heterogeneous infrared sensors
US9207708B2 (en) 2010-04-23 2015-12-08 Flir Systems, Inc. Abnormal clock rate detection in imaging sensor arrays
US9848134B2 (en) 2010-04-23 2017-12-19 Flir Systems, Inc. Infrared imager with integrated metal layers
US9143703B2 (en) 2011-06-10 2015-09-22 Flir Systems, Inc. Infrared camera calibration techniques
US10169666B2 (en) 2011-06-10 2019-01-01 Flir Systems, Inc. Image-assisted remote control vehicle systems and methods
US9900526B2 (en) 2011-06-10 2018-02-20 Flir Systems, Inc. Techniques to compensate for calibration drifts in infrared imaging devices
WO2012170954A2 (en) 2011-06-10 2012-12-13 Flir Systems, Inc. Line based image processing and flexible memory system
US10841508B2 (en) 2011-06-10 2020-11-17 Flir Systems, Inc. Electrical cabinet infrared monitor systems and methods
CN109618084B (en) 2011-06-10 2021-03-05 菲力尔系统公司 Infrared imaging system and method
US10389953B2 (en) 2011-06-10 2019-08-20 Flir Systems, Inc. Infrared imaging device having a shutter
US9961277B2 (en) 2011-06-10 2018-05-01 Flir Systems, Inc. Infrared focal plane array heat spreaders
US10051210B2 (en) 2011-06-10 2018-08-14 Flir Systems, Inc. Infrared detector array with selectable pixel binning systems and methods
US10079982B2 (en) 2011-06-10 2018-09-18 Flir Systems, Inc. Determination of an absolute radiometric value using blocked infrared sensors
US9509924B2 (en) 2011-06-10 2016-11-29 Flir Systems, Inc. Wearable apparatus with integrated infrared imaging module
US9235023B2 (en) 2011-06-10 2016-01-12 Flir Systems, Inc. Variable lens sleeve spacer
US9706137B2 (en) 2011-06-10 2017-07-11 Flir Systems, Inc. Electrical cabinet infrared monitor
US9058653B1 (en) 2011-06-10 2015-06-16 Flir Systems, Inc. Alignment of visible light sources based on thermal images
US9811884B2 (en) 2012-07-16 2017-11-07 Flir Systems, Inc. Methods and systems for suppressing atmospheric turbulence in images
US9973692B2 (en) 2013-10-03 2018-05-15 Flir Systems, Inc. Situational awareness by compressed display of panoramic views
US11297264B2 (en) 2014-01-05 2022-04-05 Teledyne Fur, Llc Device attachment with dual band imaging sensor
CN109376693A (en) * 2018-11-22 2019-02-22 四川长虹电器股份有限公司 Method for detecting human face and system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20010031073A1 (en) * 2000-03-31 2001-10-18 Johji Tajima Face recognition method, recording medium thereof and face recognition device
US6539352B1 (en) * 1996-11-22 2003-03-25 Manish Sharma Subword-based speaker verification with multiple-classifier score fusion weight and threshold adaptation
US20030063780A1 (en) * 2001-09-28 2003-04-03 Koninklijke Philips Electronics N.V. System and method of face recognition using proportions of learned model
KR20060063599A (en) * 2004-12-07 2006-06-12 한국전자통신연구원 User recognition system and that method

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3530363B2 (en) 1997-11-26 2004-05-24 日本電信電話株式会社 Recognition model generation method and image recognition method
JP4121808B2 (en) 2002-08-29 2008-07-23 三菱電機株式会社 Dictionary compression device

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6539352B1 (en) * 1996-11-22 2003-03-25 Manish Sharma Subword-based speaker verification with multiple-classifier score fusion weight and threshold adaptation
US20010031073A1 (en) * 2000-03-31 2001-10-18 Johji Tajima Face recognition method, recording medium thereof and face recognition device
US20030063780A1 (en) * 2001-09-28 2003-04-03 Koninklijke Philips Electronics N.V. System and method of face recognition using proportions of learned model
KR20060063599A (en) * 2004-12-07 2006-06-12 한국전자통신연구원 User recognition system and that method

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102156871A (en) * 2010-02-12 2011-08-17 中国科学院自动化研究所 Image classification method based on category correlated codebook and classifier voting strategy
CN101984576A (en) * 2010-10-22 2011-03-09 北京工业大学 Method and system for authenticating anonymous identity based on face encryption
ES2432479R1 (en) * 2012-06-01 2014-05-28 Universidad De Las Palmas De Gran Canaria Method for the identification and automatic classification of arachnid species through their spider webs
CN104036254A (en) * 2014-06-20 2014-09-10 成都凯智科技有限公司 Face recognition method
DE102015200434A1 (en) * 2015-01-14 2016-07-14 Bayerische Motoren Werke Aktiengesellschaft Method and apparatus for improving object recognition in different lighting situations
DE102015200433A1 (en) * 2015-01-14 2016-07-14 Bayerische Motoren Werke Aktiengesellschaft Method and apparatus for reducing the testing effort in the evaluation of an object recognition system
DE102015200437A1 (en) * 2015-01-14 2016-07-14 Bayerische Motoren Werke Aktiengesellschaft Method and device for determining the confidence of an object recognition
CN109977803A (en) * 2019-03-07 2019-07-05 北京超维度计算科技有限公司 A kind of face identification method based on Kmeans supervised learning

Also Published As

Publication number Publication date
KR100870724B1 (en) 2008-11-27

Similar Documents

Publication Publication Date Title
WO2008147039A1 (en) System and method for recognizing images using t-test
US8787629B2 (en) Image processing based on line-of-sight of a person
JP5174445B2 (en) Computer-implemented video scene boundary detection method
US8880566B2 (en) Assembler and method thereof for generating a complex signature of an input multimedia data element
US10706549B2 (en) Iterative method for salient foreground detection and multi-object segmentation
US20200151434A1 (en) Face image retrieval methods and systems, photographing apparatuses, and computer storage media
US9773189B2 (en) Recognition apparatus and recognition method
US8150169B2 (en) System and method for object clustering and identification in video
JP5385759B2 (en) Image processing apparatus and image processing method
JP6448325B2 (en) Image processing apparatus, image processing method, and program
WO2015180042A1 (en) Learning deep face representation
US9202109B2 (en) Method, apparatus and computer readable recording medium for detecting a location of a face feature point using an Adaboost learning algorithm
JP4098021B2 (en) Scene identification method, apparatus, and program
US9147130B2 (en) Information processing method, information processing apparatus, and recording medium for identifying a class of an object by using a plurality of discriminators
Willamowski et al. Probabilistic automatic red eye detection and correction
US9299008B2 (en) Unsupervised adaptation method and automatic image classification method applying the same
CN115311262A (en) Printed circuit board defect identification method
CN102722732B (en) Image set matching method based on data second order static modeling
JP5755046B2 (en) Image recognition apparatus, image recognition method, and program
Yi et al. Long-range hand gesture recognition with joint ssd network
US7734096B2 (en) Method and device for discriminating obscene video using time-based feature value
US20120201470A1 (en) Recognition of objects
JP4447602B2 (en) Signal detection method, signal detection system, signal detection processing program, and recording medium recording the program
JP2006133941A (en) Image processing device, image processing method, image processing program, and portable terminal
US11113580B2 (en) Image classification system and method

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 08723701

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 08723701

Country of ref document: EP

Kind code of ref document: A1