阅读说明：本技术 一种基于机器视觉的活体鳗鱼重量分级分选方法及装置 (Weight grading and sorting method and device for live eels based on machine vision ) 是由 闫国琦 陈尔怡 刘桥辉 莫嘉嗣 李钿 于 2019-11-08 设计创作，主要内容包括：本发明公开了一种基于机器视觉的活体鳗鱼重量分级分选方法及装置,该装置包括运送系统、拍照测重系统和分选系统；所述运送系统包括料槽和滑槽,所述拍照测重系统包括透光滑板、相机、相机支架、背光源和处理器,所述分选系统包括分拨器、分拨器驱动电机、分选箱体和传送带。本发明可以较为准确地估算鳗鱼重量,提高鳗鱼重量分级分选的工作效率,代替人工估重,减少劳动成本。(The invention discloses a weight grading and sorting method and a weight grading and sorting device for live eels based on machine vision, wherein the device comprises a conveying system, a photographing and weight measuring system and a sorting system; the conveying system comprises a trough and a sliding chute, the photographing and weight measuring system comprises a light-transmitting sliding plate, a camera support, a backlight source and a processor, and the sorting system comprises a distributor, a distributor driving motor, a sorting box body and a conveying belt. The invention can accurately estimate the weight of the eels, improves the working efficiency of classifying and sorting the weight of the eels, replaces manual weight estimation and reduces the labor cost.)

1. The utility model provides a live body eel weight classification sorting unit based on machine vision which characterized in that: the system comprises a conveying system, a photographing and weight measuring system and a sorting system; the conveying system comprises a material groove and a sliding groove, the opening of the material groove is fixedly connected with the upper end of the sliding groove in the same horizontal direction, and the lower end of the sliding groove is fixedly connected with the light-transmitting sliding plate; the photographing weight measuring system comprises a light-transmitting sliding plate, a camera support, a backlight source and a processor, wherein the camera is fixed on the camera support, a lens of the camera faces the upper surface of the light-transmitting sliding plate, the backlight source is positioned below the light-transmitting sliding plate, and the camera and the backlight source are respectively connected with the processor through leads; the sorting system comprises a distributor, a distributor driving motor, a sorting box body and a conveyor belt, wherein the upper surface of the conveyor belt and the upper surface of the light-transmitting sliding plate are at the same horizontal height; the outside of each minute dialler corresponds a separation box, selects separately box fixed mounting on the support frame of conveyer belt.

2. The live eel weight grading and sorting device of claim, wherein: the backlight source comprises a light-transmitting waterproof plate, a light mixing plate and an LED array white light source; the LED array white light source is arranged on the bottom layer and emits light to provide a light source when in work; the light mixing plate is arranged above the LED array white light source, and the white light emitted by the LED array white light source is filtered and mixed into uniform white light; the light-transmitting waterproof plate is arranged above the light mixing plate, and plays a role in waterproof insulation.

3. A weight grading and sorting method for live eels based on machine vision is characterized by comprising the following steps: the live eel weight grading and sorting device of claim 1 or 2 is adopted, and comprises the following steps:

(1) the method comprises the following steps of pouring live eels in batches into a trough, enabling the eels to freely slide down through a sliding chute one by one, triggering a camera and capturing a frame of static image after the eels pass through a light-transmitting sliding plate, and reducing the influence of reflection and shadow caused by natural light on calculation weight error because a backlight source provides a backlight source;

(2) the live eel static image is transmitted to a processor, the processor preprocesses the static image to obtain an eel contour line graph, and the eel weight is estimated according to an eel area/length and weight mathematical model;

(3) the weight grading quantity and range are preset in the processor, then the time of the eel in front of each distributor is calculated according to the level corresponding to the weight of a single live eel and the speed of the conveyor belt, the action of each distributor is controlled by a time parameter method, the eels on the conveyor belt are distributed into the corresponding sorting boxes, and the eel weight sorting and grading are completed.

4. The method of claim 3, wherein the live eel is selected by weight: the eel area/length and weight mathematical model comprises an eel body front part area and weight mathematical model and an eel body rear part middle line length and weight mathematical model, and comprises the following concrete steps:

(1) processing a frame of live eel still picture by a program to obtain an eel contour line map, which comprises the following steps: contour line, front midline, rear midline, proportional dividing line, head and tail;

(2) identifying the head and the tail of the eel by a machine learning algorithm; the top end of the head part and the top end of the tail part are connected through a center line, the distance from a perpendicular line of any point on the center line to the point is equal to that from two intersection points on the contour line, the center line is divided into a front center line and a rear center line by a proportional dividing line, and the value range of the proportional dividing line is 0.7-0.8;

(3) dividing the front part and the rear part of the eel body by a proportional dividing line; the front parts of the eel bodies with equal weight are different in thickness, the weight of the eel is measured and calculated by adopting the area, the area of the front part of the eel body, which is surrounded by the contour line on one side of the head of the eel and the proportional dividing line, is x, and a mathematical model G of the area of the front part of the eel body and the weight is established according to the eel dividing experiment₁＝F₁(x) (ii) a The back part of eel body is flat and has uncontrollable posture, the weight of eel is measured by adopting length, the length of the middle line of the back part of eel body is y, and the eel body is established according to eel segmentation experimentMid-line length and weight mathematical model G₂＝F₂(y)；

(4) The mathematical model of eel area/length and weight is as follows: g₃＝a·F₁(x)+b·F₂(y); wherein a is an area weight coefficient, and the value range of a is 0.7-0.8; b is a length weight coefficient, and the value range of b is 0.2-0.3; g₃Is the actual weight of eel.

Technical Field

The invention relates to the field of fish grading and sorting equipment, in particular to a live eel weight grading and sorting method and device based on machine vision.

Background

The eel is a fish with the appearance similar to a long and snakelike shape, has no scales, has a cylindrical trunk part, a slightly flat tail part, developed mucous glands and smooth body surface, is generally produced in a salt and fresh water boundary sea area, and has the migration characteristic similar to salmon. The weight selection of the artificially cultured eels mainly adopts an artificial selection mode, the eels are fished out of a feeding pond, the activity of the eels is reduced through low-temperature treatment, and the eels are classified and selected by manually estimating the weight; however, the manual sorting method has low accuracy and low efficiency, and the low-temperature treatment affects the activity of the eels. At present, the automatic fish weight grading and sorting equipment has higher cost, generally utilizes a motion weighing principle, namely a weighing device weighs a conveying part and a weighing object together, and the weighing mode has low precision and large interference and is not suitable for live eels; on the other hand, the activity of live eel is stronger, and the eel self-motion also can influence the accuracy of weighing when passing through weighing device. Therefore, the development of a weight grading and sorting method and equipment suitable for the characteristics of live eels is urgently needed.

Disclosure of Invention

The invention aims to overcome the defects in the prior art, and provides a live eel weight grading and sorting method and device based on machine vision, which can quickly measure and grade eels, have high accuracy and high efficiency, effectively reduce labor cost, and simultaneously can reduce secondary damage to the eels during manual sorting.

The purpose of the invention is realized by the following technical scheme:

a live eel weight grading and sorting device based on machine vision comprises a conveying system, a photographing and weight measuring system and a sorting system; the conveying system comprises a material groove 1 and a sliding groove 2, wherein the opening of the material groove 1 is fixedly connected with the upper end of the sliding groove 2 in the same horizontal direction, and the lower end of the sliding groove 2 is fixedly connected with a light-transmitting sliding plate 3; the photographing weight measuring system comprises a light-transmitting sliding plate 3, a camera 5, a camera support 4, a backlight source 6 and a processor 7, wherein the camera 5 is fixed on the camera support 4, a lens of the camera 5 faces the upper surface of the light-transmitting sliding plate 3, the backlight source 6 is positioned below the light-transmitting sliding plate 3, the camera 5 and the backlight source 6 are respectively connected with the processor 7 through leads, and the processor 7 is used for data processing and man-machine interaction; the sorting system comprises a distributor 8, a distributor driving motor 20, a sorting box 9 and a conveyor belt 10, wherein the upper surface of the conveyor belt 10 and the upper surface of the light-transmitting sliding plate 3 are at the same horizontal height, the distributor driving motor 20 is fixedly arranged on a support frame of the conveyor belt 10, and the distributor 8 is arranged on the distributor driving motor 20; the outer side of each distributor 8 corresponds to one sorting box body 9, and the sorting box bodies 9 are fixedly arranged on a supporting frame of a conveying belt 10.

The backlight source 6 comprises a light-transmitting waterproof plate 17, a light-mixing plate 18 and an LED array white light source 19; the LED array white light source 19 is arranged on the bottom layer and emits light to provide a light source when in work; the light mixing plate 18 is arranged above the LED array white light source 19, and the white light emitted by the LED array white light source 19 is filtered and mixed into uniform white light; the light-transmitting waterproof plate 17 is arranged above the light mixing plate 18 and plays a role in waterproof and insulation.

A weight grading and sorting method of live eels based on machine vision adopts the weight grading and sorting device of live eels based on machine vision, and comprises the following steps:

(1) the method comprises the following steps of pouring live eels in batches into a trough 1, manually operating to enable the eels to freely slide down through a chute 2 one by one, enabling the eels to pass through a light-transmitting sliding plate 3, triggering a camera 5 and capturing a frame of static image, and providing a backlight source 6 for a backlight source to reduce the influence of reflection and shadow caused by natural light on calculated weight errors;

(2) the live eel static image is transmitted to a processor 7, the processor 7 preprocesses the static image to obtain an eel contour line graph, and the eel weight is estimated according to an eel area/length and weight mathematical model;

(3) the weight grading quantity and range are preset in the processor 7, then the time of the eel arriving at the front of each distributor 8 is calculated according to the level corresponding to the weight of a single live eel and the speed of the conveyor belt 10, each distributor 8 is controlled to act by a time parameter method, the eels on the conveyor belt 10 are distributed into the corresponding sorting box 9, and the eel weight grading is completed.

The eel area/length and weight mathematical model comprises an eel body front part area and weight mathematical model and an eel body rear part middle line length and weight mathematical model, and comprises the following concrete steps:

(1) processing a frame of live eel still picture by a program to obtain an eel contour line map, which comprises the following steps: contour line 11, front midline 12, rear midline 13, proportional dividing line 14, head 15, tail 16;

(2) identifying the head 15 and the tail 16 of the eel by a machine learning algorithm; the top end of the head part 15 and the top end of the tail part 16 are connected through a center line, the distance from a perpendicular line of any point on the center line to the point is equal to that from two intersection points on the contour line 11, the center line is divided into a front center line 12 and a rear center line 13 by a proportional dividing line 14, and the value range of the proportional dividing line 14 is 0.7-0.8;

(3) dividing the front part and the rear part of the eel body by a proportional dividing line 14; the front parts of the eel bodies with equal weight are different in thickness, the weight of the eel is measured by adopting the area, the area of the front part of the eel body, which is enclosed by the contour line 11 on one side of the eel head 15 and the proportional dividing line 14, is x, and a mathematical model G of the area of the front part of the eel body and the weight is established according to an eel division experiment₁＝F₁(x) (ii) a The back part of the eel body is flat and the posture is not controllable, the weight of the eel is measured by adopting the length, the length of the middle line of the back part of the eel body is y, and a mathematical model G of the length and the weight of the middle line of the back part of the eel body is established according to an eel segmentation experiment₂＝F₂(y)；

(4) The mathematical model of eel area/length and weight is as follows: g₃＝a·F₁(x)+b·F₂(y); wherein a is an area weight coefficient, and the value range of a is 0.7-0.8; b is a length weight coefficient, and the value range of b is 0.2-0.3; g₃Is the actual weight of eel.

Compared with the prior art, the invention has the following advantages and effects:

(1) the method adopts a machine vision weight measuring technology, and the front part of the eel body is cylindrical and has different thicknesses; the invention utilizes the machine vision technology to process the static image captured by the live eel, divides the live eel into a plurality of sections according to the length proportion, and estimates the weight of the eel by adopting a method of combining the area and the length, thereby having high precision and low cost.

(2) The device of the invention can estimate the weight of the eel more accurately, improve the working efficiency of classifying and sorting the weight of the eel, replace manual weight estimation and reduce the labor cost.

Drawings

Fig. 1 is a perspective view of a machine vision-based live eel weight grading and sorting device.

Fig. 2 is a front view of the structure of the live eel weight grading and sorting equipment based on machine vision.

Fig. 3 is a structural top view of the live eel weight grading and sorting equipment based on machine vision.

Fig. 4 is a schematic structural diagram of the photographing weight-measuring system.

Fig. 5 is a schematic structural diagram of a backlight.

Figure 6 is an eel contour line diagram.

Wherein, 1, a trough; 2. a chute; 3. a light-transmitting slide plate; 4. a camera support; 5. a camera; 6. a backlight source; 7. a processor; 8. a distributor; 9. a sorting box body; 10. a conveyor belt; 11. contour lines; 12. an anterior midline; 13. a posterior midline; 14. dividing the line according to proportion; 15. a head portion; 16. a tail portion; 17. a light-transmitting waterproof sheet; 18. a light mixing plate; 19. an LED array white light source; 20. the distributor drives the motor.

Detailed Description

The present invention will be described in further detail with reference to examples, but the embodiments of the present invention are not limited thereto.