CA1251863A

CA1251863A - Fish sorting machine

Info

Publication number: CA1251863A
Application number: CA000560139A
Authority: CA
Inventors: Kevin Mccarthy; Patrick Debourke
Original assignee: Grove Telecommunications Ltd
Current assignee: Grove Telecommunications Ltd
Priority date: 1988-02-29
Filing date: 1988-02-29
Publication date: 1989-03-28
Also published as: DK93389A; DK93389D0; US4963035A; NO167182B; EP0331390A3; NO890853D0; NO890853L; EP0331390A2

Abstract

ABSTRACT
Means for sorting fish comprising means for receiving the image of a fish in a video camera, means for receiving the image from the camera and for storing it in an image plane memory, means for determining boundary values of the image, means for determining predetermined areas, length and width of the image, means for determining ratios of predetermined ones of the areas, length and/or width, and means for indicating the specie of the fish from predetermined ratios of the areas.

Description

~;~5~

01 This inventlon relates to a fish 02 processing system and in particular to a novel fish 03 classification method and system.
04 Fish processing plants traditionally have 05 soLted ish by the use o-E a human operator to 06 determine various characteristics of fish, for example 07 the size, weight, sex, specie of fish, etc. The fish 08 are typically sorted into various chutes for 09 collection by group and later processing. In recent years automatic means have been designed for 11 classifying the fish whereby they can be sorted. For 12 example in Canadian Patent 1,039,235 it is desired to 13 determine whether a fish is a female carrying eggs.
14 Means is provided for automa-t-ically determining the transparency or translucency of the more light 16 transparent female fish carrying eggs, and on the 17 basis of this determination, those fish are routed 18 along an unique chute. In that patent it is important 19 to orient the fish head forward, and an apparatus described therein performs the orienting func-tion.
21 Other apparatus for processing fish is 22 described in U.5. patent 4,339,588 issued 23 August 23, 1983 to Robert ~olnar. Patents dealing 24 with automatic determination of the characteris-tics of certain items are described in Canadian Patent 26 1,110,996 and U.S. Patents 4,687,107, 4,630,225, 27 4,324,335, and 4,351,437.
28 None of the above patents have the ability 29 to determine the specie of the fish being considered.
Further, none have the ability of processing several 31 rows of fish at the same time. In addition, none can 32 determine whether the fish is 'nead -Eirst or tail 33 first, estimate the weight of the Eish, etc.
34 In prior art systems where the length of an object is to be determined, the front to back 36 length is determined. However where fish are 37 concerned, it is possible that the fish is curved, and ~L
q~

~:~S~ 3 01 that t~e front to back distance does not represent 02 with any accuracy the length of the fish. ~ccordingly 03 if a weight determination ;s ~o be made based on the 04 measured front to rear length, it will be inaccurate, 05 especially since t~e area is unknown, and the specie, 06 which can have a different weight per unit volume are 07 from the other.
08 The present invention is a means which 09 provides a specie indication of the fish. It also provides means for providing the true length of the 11 fish, whether the fish is oriented tail first or head 12 first on a conveyor. Further, the i.nvention includes 13 means for providing the above where there are multiple 14 rows of fish coming along a conveyor belt at the same time, using only a single camera.
16 Once an indication has been made as to the 17 specie, size, various aspect ratios of the fish, they 18 can be separated by means of pneumatically operated 19 doors disposed across the conveyor belt. The operation of the doors is not the subject of the 21 present in~ention.
22 In accordance with one aspect of the 23 invention, a method for sor-ting Eish is comprised of 24 establishing the digital values of the bit plane pixel positions at the edges of the image of a fish stored 26 on a bit plane, along lines across the fish, 27 determining the midpoints between the values, and 28 summing the distances between the midpoints, to obtain 29 a digital value signal representative of the length of the fish.
31 In accordance with another aspect of the 32 invention, a method for sorting fish is comprised o-f 33 the further steps of determining the midpoint of the 34 summed distance, establishing the sum of the distance between the bit plane pixel positions at the edge of 36 the image to the lef-t of the midpoint of the summed 37 distances b~tween the midpoint to ob-tain a left area ~S~63 01 signal, establishing the sum o the distance between 02 the bit plane pixel positions at the edge of the image 03 to the right of the midpoint oE the summed distances 04 between the midpoints to obtain a right area signal, 05 comparing the left area signal with the right area 06 signal and providing a signal indicative of the fish 07 being positioned tail -to the right in the event the 08 left area signal is equal to or smaller than the right 09 area signal.
In accordance with a further aspect, the 11 invention is comprised of means for receiving the 12 image of a fish in a video camera, means for .r~ceiving 13 the image from the camera and for storing it in an 14 image plane memory, means -for determining boundary values of the image, means for determining 16 predetermined areas, length and width of the image, 17 means for determining ratios of predetermined ones of 18 the areas, length and/or width, and means for 19 indicating the specie of the fish from prede-termined ratios of the areas.
21 In accordance wit'n another aspect, the 22 invention is comprised of means for receiving the 23 image of a fish in a video camera, means for receiving

2~ the image from the camera and for storing it in an image plane memory, means -for determining stored 2~ values of the width of the fish along lines 27 perpendicular to the principal axis of the fish, and 28 stored values of the position of the first and last 29 one of the lines along the principal axis which intersect the image, means -for determining the 31 positions of the midpoints between the determined 32 width boundary values, and for determining the length 33 of lines joining the midpoints, means for summing the 34 leng-th of the lines to determine the length of the 35 fish.
36 A better understanding of the invention 37 will be better understood by re-ference to the detailed 81~3 01 description below, in conjunction with the following 02 drawings, in which:
03 Figure 1 is a plan view of a conveyor belt 04 conveying two rows of fish, 05 Figure 2 is a front elevation of the view 06 shown in Figure 1, 07 Figure 3 is a block diagram of the present 08 invention, 09 Figure 4 is a schematic representation of a memory used in the processor of Figure 3, 11 Figure 5 is a view of the bit plane used 12 to illustrate how the dimensions of a fish are 13 determined, 14 Figure 6 is a view of a bit plane illustrating how the true length of a fish is 16 determined, 17 Figures 7~-7D is a flow chart of an 18 algorithm for determining the position of the Eish 19 represen-ted on a memory bit plane connected to a video camera, and for determining the specie of the fish, 21 and 22 Figures 8A-8C represent a flow chart of an 23 algorithm for determining the presence of a fish in 24 the bit plane.
Figure 1 illustrates a conveyor 1, moving 26 in the direction of the arrow, carrying four fish 27 2A-2D of two lines of fish in two rows separated by a 28 dotted line along the axis of the arrow. ~t a 29 position downstream of the fish and disposed across either a further conveyor or a chute 3 are preferably 31 pneumatically controlled doors 4 which can take either 32 the position shown or an alternate position depicted 33 in phantom. When the doors are in the positions 34 shown, wha-tever fish pass thereto along the chutes 3, are deflected into a lower sub-chu-tes 5, for further 36 processing selectively relating to that sub-chute. As 37 may be seen either one of the chutes has two associate ~25~363~

0l sub-chutes, and each sub-chu-te can be further 02 divided~ Thus with the knowledge oE what specles, 03 weight, or other characteristic is associated with the 04 fishl a signal can be generated and with time delay 05 sent to a controlling mechanism for one or both of the 06 doors, to cause -the fish to pass down one of the 07 sub-chutes as desired. It can be seen in Figure 1 08 that fish 2D is tail first. The present invention is 09 able to handle the fish determination with the fish tail first in contrast to the prior art which required 11 it to be head first.
12 Figure 2 depicts the apparatus of Figure 1 13 in a side elevation. In this figure a video camera 6 14 is diposed above the conveyor, by which a field of view spanning across both rows of fish is registered.
16 It is preferred that video camera 6 should have a 17 charge coupled device (CCD) light sensor. One camera 18 which can be used in the present invention is ~ade by 19 Hitachi Corp., type KP120 CCD camera.
A block diagram of apparatus for carrying 21 out the process of the present invention is shown in 22 Figure 3. Video camera 6 is connected to a video or 23 image processor 7, ~hich provides an output signal 24 display and/or modem 8 and/or printer 8A, and controls signals to controller 9. The output signal of 26 controller 9 is connected to chute door control 27 apparatus, shown as chute solenoids l0. The chute 28 solenoids are connected to the doors or to pneumatic 29 controls for pneumatically opera-ting doors 4O The image processor is preferred to be OCTEK 2000 Viking 31 task, operated by a type ll/73 microcomputer of 32 Digital E~uipment Corporation.
33 The video camera in conjunction with 34 processor 7 generates four bit words which are characteristic of each pixel containing grey tone in 36 the CCD of the video camera 6. This signal is 37 normally stored in a bit plane. For example the video 38 memory of the sys-tem typically consists of 320 x 240 ~lt ~25~L~63 Ql fo~lr bit words. This allows storage of one -full image 02 usin~ sixteen grey levels. However in the present 03 invention we consider the memory as consisting of four 04 groups of 320 x 240 one bit words, contained in each 05 of four bit planes. Thus four full binary images are 06 stored therein. With a binary image, each pixel is 07 either black or white, i.e. on or off. Thus only one 08 bit, e.g. the most significant bit of each word is 09 present on the first bit plane.
In the present invention the bit plane ll containing the most significant bit is utilized. The 12 data to be stored in the other three bit planes is 13 filtered or masked.
14 Considering Figure 4, four bit planes are shown into which the image memory is typically 16 divided. In the present invention, however, the most 17 significant bit is stored in the top most bit plane 18 11. The other bit planes are used as will be l9 described below.
The bit plane ll is continuously scanned.
21 As the image of a fish 12 progresses from right to 22 left across the bit plane, it eventually crosses a 23 line starting at the left of the bit plane at position 24 x=40, y=10. This is s~nsed once the memory locations along that line change from 0 to 1 or vice versa, in 26 the presence of a fish.
27 It should be noted that there are two 28 lines oE fish, separated by the imaginary dashed line 29 13. The processor divides the bit plane into two, on each side of line 13, and processes the bit plane 31 image therein separately. Thus even if two fish pass 32 the line 0,0 separately, the processor deals with that 33 information individually.
34 With the indication of a fish having passed line 40,10 in the left hand row, which has not 36 passed line lO,lO (which is close to the end of the 37 bit plane), the image data from the left-hand row is ~:5~8~
01 transferred -to the second bit plane 14. In the same 02 manner when the presence of a fish 12A is present in 03 the right-hand row which has passed line 40,10 but has 04 not yet passed line 10,10 is detected, the bit plane 05 image from that side of line 30 is transferred to the 06 third bit plane 15.
07 Thus because the -fish can overlap each 08 other and indeed be located exactly coincident with 09 each other in respect of a line or-thogonal to the direction of travel, yet the time between fish along a 11 line constituting a row is relatively long, the 12 present invention can provide sufficient processing 13 time to accommodate more than one row of fish by 14 transferring the ~ish image from each of the rows -to individual separate bit planes where they re stored 16 for longer periods of time. In a similar manner, for 17 example, three rows of fish can be accommodated, and 18 the image of the third row transferred to the fourth 19 bit plane 16, etc.
Once the images have been transEerred to 21 bit planes other than the first, the processor can 22 deal with each in leisure. Further, the images can be 23 transferred to a C~T display either automatically or 24 selectively by an operator. The specie and/or other factors determined about each fish can be transferred 26 to a remote computer by means of a modem 8 and printed 27 on a printer 8A.
28 Figure 5 illustrates the outline 17 of a 29 grey image of a fish on one of the lower bit planes.
The bit plane is digitally scanned, the la-teral scan 31 lines being depicted as 171, 172, 173... 17~_3, 32 17N-2l 17N-1l 17N- It can be seen that scanning 33 lines 171 and 17~ all encounter lines of whi-te 34 pixels, e.g. 0's. Elowever immediately inward scan lines 172 and 17N_l depict lines in which black 36 pixels (containing l's) are encountered.
37 For the lines containing black pixels~ the 16~l 01 digital values at the minimum and maximum digital 02 values of -the black pixels are determined, and divided 03 by 2. The pixel points representing halE the distance 04 between the two aforenoted minimum and maximum black 05 pixels are depicted as center points 182, 06 183---18N-1~ 18N-2- 18N-3 (the reference numeral 07 subscripts of the center points where chosen to 08 correspond with the corresponding subscript number for 09 the scanning lines, for sake of consistency).
Figure 6 depicts the fish image 17 with 11 the scanning lines 17~_1 as in Figure 5. In this 12 case the fish image 17 is curved, corresponding to a 13 curved fish. The center points 182-18N_l are also 14 shown.
Now the distances between the center 16 points are all determined and summed. This sum 17 represents the overall length of the fish, whether 18 straight or curved. In contrast to this, prior ar-t 19 imaging systems would determine the head to tail distance represented by line 19. It can be seen that 21 the length of the line 19 is substantially shorter 22 than the length of the line representing the distances 23 between the center points. The latter line, which 24 follows the curvature of the fish, is clearly the more accurate representation.
26 The length of the fish having been 27 de-termined, the half length point 18H is then 28 determined, by determining the scan line which 29 intersec-ts the center point which is half the determined distance between the end point center 31 points 182 and 18N_~. The distance 20 represents 32 the left half of the fish images and the distance 21 33 represents the right half of the fish image.
34 With the addresses of the first black pixels along each scan line representing the edges of 36 the fish image already determined, the sum of the 37 distances between them (i.e. representing the widths ~5~L8~;3 01 of the fish along those scan lines) is determined Eor 02 both the left half of the fish and the righ-t half of 03 the fish, along the scan lines intersecting the lines 04 20 and 21 respectively. This corresponds to 05 determining areas of the image surface. A comparison 06 is then made. If the sum of the widths of the scan 07 lines intersecting the fish image over the lef- half 08 of the fish is equal to or greater than the sum of the 09 scan line widths intersecting the image at the right half of the fish, then the fish is considered to be 11 nose forward (to the left). However if the sum of the 12 left half of the fish is not equal to or greater than 13 the sum of the right half of the fish, then the fish 14 is considered to be position tail forward. In order to determine the remainer of the values, the elements 16 in all arrays relating to this tail forward fish image 17 are merely reversed. Thus the requirement to 18 physically turn the fish 180 nose to end as required 19 in the prior art is not required in the present invention.
21 The aspect ratio (length/width), area 22 ratio (area/length), tail ratio (width of tail/square 23 root of length), head ratio (width of head/square root 24 of length), nose angle and symmetry of the fish are determined from the image edge positions and 26 calculated values. In the event the aspect ratio is 27 greater than 5.5 and the area ratio is smaller than 28 30.0, it is concluded that a codfish is present.
29 With the aspect ratio greater than 5.5 and the area ratio greater than 30.0, it is determined 31 that a catfish is present.
32 In the event the aspect ratio is smaller 33 than 5.5 and the area ratio is smaller than 30.0, it 34 is determined that a redfish is present. In the event the aspect ratio is smaller than 5.5 and the area 36 ratio is greater than 30.0, it is concluded that a 37 flatfish is present.
38 _ 9 _ ~5~3 01 With the determina-tion o t~e aforenoted 02 four species, Elags are set appropriate to each of the 03 sp~cies.
04 With the area being determined by summing 05 the widths of each of the scan lines over the interval 06 which intersect the dark pixels in the bit plane, the 07 area is determined. With the area calculated and 08 in the presence of a flag indicating the fish specie, 09 a constant is applied appropriate to each specie and an assumed weight is calculated.
11 On the basis of the weight and/or specie, 12 the processor 7 provides a signal to the controller 9 13 indicative of the designation of the fish resident on 14 the particular bit plane being processed. The controller, with an appropriate time delay relating to 16 the speed of the conveyor, generates a signal to chute 17 solenoids 10 to control doors 4 and the ultimate 18 destination of the fish.
19 The above description has followed the processing of the image along the general algori-thm 21 described in Figures 7A-7D respectively. The flow 22 charts shown therein, which will be evident to a 23 person skilled in the art and containing descriptions 24 of the process, are an indication of the process in more detail. The flow charts consti-tuting Figures 8A, 26 8B and 8C represent the interrupt routine for 27 determining the presence of a fish on -the first image 28 plane and the determination of its dimensions on the 29 second or third, as described more generally above.
Because considerable processing time is 31 required to do the specie and weight determination and 32 to allow for display on a monitor, while the 33 determination of the presence of fish in the two lanes 34 in the first bit plane can be done rapidly, the present invention has provided means for sharing a 36 processor to perform the various tas~s in a highly 37 efficient manner, and to perform determination of fish 01 characteristics not previously able -to be provided.
02 As noted earlier the determination of -the presence of 03 a fish is performed in association with the leas-t 04 significant bit in the bit image constituting a single 05 bit plane, and once the presence oE the fish image has 06 been determined on either side of an imaginary line 07 dividing the fish row paths, the entire image from 08 either half is stored on corresponding separate bit 09 planes. The processor can, after determining the presence of the image and transferring it to one of 11 the other bit planes, perform the specie determination 12 calculations with more available time, while being 13 interrupted from time to time to check for the 14 presence of a fish in the first bit plane.
Further, the presen-t inven-tion determines 16 the length of the fish, including curved fish, to a 17 much more accurate degree than previously possible.
18 This is performed by calculating the midpoint of the 19 scanning lines over the image of the fish, then summing the distances between the midpoints.
21 With the availability of the midpoint 22 values, the center of the fish coordinates are 23 determined, and the areas oE the fish to the left and 24 right determined. This provides an indication of whether the fish is tail or head first.
26 ~ith the summation of the scan line 27 distances over the fish image, the area is determined, 28 and from that a specie calculation as described in the 29 body of -this specification, and an estimated weight of the fish can be determined. With an appropriate 31 resulting signal being sent to a chute solenoid 32 controller, the fish can be sorted to an accurate 33 degree and with characteristic sub-divisions not 34 previously possible.
A person skilled in the art understanding 36 this invention may now conceive of alternate 37 embodiments or other designs using the principles ~5~8~i~

01 described herein. All are considered to be within the 02 sphere and scope of this invention as defined in the 03 claims appended here-to.

Claims

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. Means for sorting fish comprising:
(a) means for receiving the image of a fish in a video camera, (b) means for receiving the image from the camera and for storing it in an image plane memory, (c) means for determining boundary values of said image, (d) means for determining predetermined areas, length and width of said image, (e) means for determining ratios of predetermined ones of said areas, length and/or width, and (f) means for indicating the specie of the fish from predetermined ratios of said areas.

2. Means for sorting fish comprising:
(a) means for receiving the image of a fish in a video camera, (b) means for receiving the image from the camera and for storing it in an image plane memory, (c) means for determining stored values of the width of the fish along lines perpendicular to the principal axis of the fish, and stored values of the position of the first and last one of said lines along said principal axis which intersect said image, (d) means for determining the positions of the midpoints between said determined width boundary values, and for determining the length of lines joining said midpoints, (e) means for summing the length of said lines to determine the length of the fish.

3. Means for sorting fish as defined in claim 2, further comprising:
(f) means for determining the midpoint of the length of the fish, and (g) means for summing the values between said determined width values to the left of the midpoint of the length of the fish to obtain a first half length area, and for summing the values between said determined width values to the right of the midpoint of the length of the fish to obtain a second half length area, and (h) means for indicating that the fish is positioned tail first if the first half length area is less than or equal to the second half length area.

4. Fish sorting means as defined in claim 3, including means for determining the aspect ratio (length/width), the area ratio (area/length), tail ratio (width of tail/square root of length) and head ratio (width of head/square root of length), means for comparing at least the aspect and area ratios with predetermined corresponding ratios, and means for providing an indicating signal corresponding to a determined fish species therefrom.

5. Means for sorting fish comprising:
(a) a conveyor belt for carrying more than one row of fish along row axes parallel to the axis of the conveyor belt, (b) means for illuminating the fish, (c) a video camera spared from the fish for receiving at least outline images of the fish carried along the belt, (d) a video memory having at least 3 bit planes connected to the video camera, (e) means for storing an image of a fish in a first row received from the video camera in a first bit plane of the video memory, (f) means for determining the presence of the image and for transferring the image of the fish to a second bit plane of the video memory, (g) means for determining the size and orientation of the fish from stored edge values of the image in the second bit plane, (h) means for storing the image of another fish in another row received from the video camera in the first bit plane of the video memory, (i) means for determining the size and orientation of the other fish from the stored edge values of the image of the other fish in the first bit plane, (j) means for transferring the image of the other fish to a third bit plane, whereby the same video memory can be used for characteristic analysis of fish carried by the conveyor belt and display by the video monitor of more than one row of said fish.

6. Means for sorting fish as defined in claim 5 further comprising:
(a) means for determining boundary values of said image in the second bit plane, (b) means for determining predetermined areas of said image, (c) means for determining ratios of predetermined ones of said areas, and (d) means for indicating the species of the fish from predetermined ratios of said areas.

7. Means for sorting fish as defined in claim 5, further comprising:
(a) means for determining stored boundary values of the width of the fish along lines perpendicular to the principal axis of the fish, and stored values of the position of the fish and first and last one of said lines along said principal axis which intersect said image, (b) means for determining the positions of the midpoints between said determined width boundary values, and for determining the length of lines joining said midpoints, and (c) means for summing the length of said lines to determine the length of the fish.

8. Means for sorting fish as defined in claim 7, further comprising:
(a) means for determining the midpoint of the length of the fish, (b) means for summing the values between said determined width values to the left of the midpoint to obtain a first half length area, and for summing the values between said determined width values to the right of the midpoint of the length of the fish to obtain a second half length area, and (c) means for indicating that the fish is positioned tail first if the first half length area is less than or equal to the second half length area.

9. Means for sorting fish as defined in claim 8, further comprising means for determining the aspect ratio (length/width), the area ratio (area/length), tail ratio (width of tail/square root of length), and head ratio (width of head/square root of length), means for comparing at least the aspect and area ratios with predetermined corresponding ratios, and means for providing an indicating signal corresponding to a determined fish species.

10. A method for sorting fish comprising:
(a) establishing the digital values of the bit plane pixel positions at the edges of the image of a fish stored on a bit plane, along lines across the fish, (b) determining the midpoints between said values, and (c) summing the distances between said midpoints, to obtain a digital value signal representative of the length of the fish.

11. A method as defined in claim 10, with the further steps of:
(d) determining the midpoint of said summed distance, (e) establishing the sum of the distance between said bit plane pixel positions at the edge of the image to the left of the midpoint of said summed distances between said midpoint to obtain a left area signal, (f) establishing the sum of the distance between said bit plane pixel positions at the edge of the image to the right of the midpoint of said summed distances between said midpoints to obtain a right area signal, (g) comparing the left area signal with the right area signal and providing a signal indicative of the fish being positioned tail to the left in the event the left area signal is equal to or smaller than the right area signal.

12. A method as defined in claim 10 or 11 including the steps of determining from said bit plane pixel positions and digital value respresentative of the length of the fish the aspect ratio (length/width), the area ratio (area/length), the tail ratio (width of tail/square root of the length) and head ratio (width of the head/square root of the length), comparing at least the aspect and area ratios with predetermined ratios, and providing a signal indicating one or more predetermined specie of fish therefrom.

13. A method as defined in claim 10 or 11 including the steps of determining from said bit plane pixel positions and digital value representative of the length of the fish, the aspect ratio (length/width), the area ratio (area/length), the tail ratio (width of tail/square root of the length) and head ratio (width of the head/square root of the length), comparing at least the aspect and area ratios with predetermined ratios, and indicating the presence of a cod fish if the aspect ratio is greater than about 5.5 and the area ratio is smaller than about 30Ø

14. A method as defined in claim 10 or 11 including the steps of determining from said bit plane pixel positions and digital value representative of the length of the fish, the aspect ratio (length/width), the area ratio (area/length), the tail ratio (width of tail/square root of the length) and head ratio (width of the head/square root of the length), comparing at least the aspect and area ratios with predetermined ratios, and indicating the presence of a catfish if the aspect ratio is greater than about 5.5 and the area ratio is greater than about 30Ø

15. A method as defined in claim 10 or 11 including the steps of determining from said bit plane pixel positions and digital value representative of the length of the fish, the aspect ratio (length/width), the area ratio (area/length), the tail ratio (width of tail/square root of the length) and head ratio (width of the head/square root of the length), comparing at least the aspect and area ratios with predetermined ratios, and indicating the presence of a redfish if the aspect ratio is smaller than about 5.5 and the area ratio is smaller than about 30Ø

16. A method as defined in claim 10 or 11 including the steps of determining from said bit plane pixel positions and digital value representative of the length of the fish the aspect ratio (length/width), the area ratio (area/length), the tail ratio (width of tail/square root of the length) and head ratio (width of the head/square root of the length), comparing at least the aspect and area ratios with predetermined ratios, and indicating the presence of a flatfish if the aspect ratio is smaller than about 5.5 and the area ratio is greater than about 30Ø

17. A method for sorting fish, including th steps of:
(a) receiving on a first bit plane of a digital memory the most significant bit values of an image of a fish as it is viewed in a video camera, (b) repeatedly checking the position of said image on the first bit plane, (c) tranferring the image to a second bit plane for analysis of its dimensions when the progressing front of the image passes a predetermined line of pixels orthogonal to the direction of progress of the image, which are in within a predetermined distance from the edge of the first bit plane.

18. A method for storing fish as defined in claim 17, further including:
(a) establishing the digital values of the second bit plane pixel positions at the edge of the image of a fish stored on a bit plane, along lines across the fish, (b) determining the midpoints between said values, and (c) summing the distance between said midpoints, to obtain a digital value signal representative of the length of the fish.

19. A method as defined in claim 18, with the further steps of:
(d) determining the midpoint of said summed distance, (e) establishing the sum of the distance between said bit plane pixel positions at the edge of the image to the left of the midpoint of said summed distances between said midpoint to obtain a left area signal, (f) establishing the sum of the distance between said bit plane pixel positions at the edge of the image to the right of the midpoint of said summed distances between said midpoints to obtain a right area signal, (g) comparing the left area signal with the right area signal and providing a signal indicative of the fish being positioned tail to the right in the event the left area signal is equal to or smaller than the right area signal.

20. A method as defined in claim 19 including the steps of determining from said bit plane pixel positions and digital value respresentative of the length of the fish the aspect ratio (length/width), the area ratio (area/length), the tail ratio (width of tail/square root of the length) and head ratio (width of the head/square root of the length), comparing at least the aspect and area ratios with predetermined ratios, and providing a signal indicating one or more predetermined specie of fish therefrom.