Automatic sorting system and method for nondestructive testing of shellfish gender

Abstract

The invention provides an automatic sorting system and method for nondestructive testing of shellfish sex, and belongs to the technical field of shellfish sex determination and sorting. According to the X-ray-based automatic sorting system for nondestructive testing of the shellfish sex, on the basis that the sex of each shellfish individual is accurately measured, male and female individuals are separately and temporarily bred through a physical means, the problem that the hatching rate is low due to excessive sperm emission caused by the large number of male individuals is effectively solved, and meanwhile the hatching efficiency is improved. The breeding quantity of male individuals can be properly reduced, the hatching rate of shellfish fry production is effectively improved, the production cost is reduced, the comprehensive benefits of shellfish fry production are improved, and technical support is provided for shellfish breeding and fry production.

Description

Technical Field

[0001] The invention belongs to the technical field of shellfish sex identification and sorting, and in particular relates to an automatic sorting system and method for non-destructive detection of shellfish sex. Background Art

[0002] my country is the world's largest shellfish farming country, and shellfish is also the most productive marine aquaculture species in my country. Shellfish farming has high value from the perspectives of protein supply, ecological restoration, job creation, and export earnings. Promoting the healthy development of the shellfish farming industry is of great significance to the economic, social and ecological development of my country's marine aquaculture industry and even coastal areas. Since the 1970s, Shandong Province has taken the lead in conducting artificial seedling research in China. Through continuous testing of artificial breeding technical parameters, it has achieved successful breakthroughs and promoted aquaculture on a large scale throughout the province and even the country.

[0003] Controlling sperm and ovulation is a key technical issue in the process of shellfish factory seeding and shellfish breeding. In traditional methods, shellfish, especially those that cannot be dissected for fertilization, make it difficult to conduct large-scale gender detection. The only way is to pick out the excessive male individuals during the sperm and egg discharge process. However, this requires high technical staff and makes it difficult to control the amount of sperm, which can easily lead to the problem of low hatching rate due to multiple sperm. Therefore, how to separate male and female shellfish in advance before fertilization and improve the controllability of the male-female ratio in shellfish seed production is an urgent problem to be solved for the shellfish seedling industry and germplasm innovation. Summary of the invention

[0004] The present invention provides an automatic sorting system and method for non-destructive detection of shellfish sex. The designed automatic sorting system can non-destructively identify the sex of shellfish with high throughput, and can separate male and female shellfish in advance before fertilization, thereby improving the controllability of the male-female ratio in shellfish seed production and comprehensively improving the technical level of shellfish seed production.

[0005] In order to achieve the above object, the present invention provides an automatic sorting system for non-destructive detection of shellfish sex, comprising the following parts:

[0006] The shellfish positioning and correction unit comprises a water conveyor belt and an automatic posture correction part, wherein the automatic posture correction part is used to correct the posture of the shellfish to be tested that are fed one by one by the water conveyor belt;

[0007] An X-ray scanning unit is connected to the shellfish positioning and correction unit through a water conveyor belt, and is used to scan the gonads of the shellfish to be tested to obtain the male and female characteristic images of the shellfish gonads;

[0008] A control unit is connected to the X-ray scanning unit through an electrical signal, and is used to control the X-ray scanning unit to perform real-time imaging and store the male and female characteristic images of the gonads of shellfish in real time;

[0009] The shellfish sex intelligent identification unit is connected to the control unit through an electrical signal, and is used to identify the sex characteristic images of the shellfish gonads stored in the control unit, and feed back the shellfish sex identification results to the automatic sorting unit;

[0010] The automatic sorting unit is connected to the control unit through electrical signals, and is used to receive the male and female discrimination results of the shellfish, and automatically sort the shellfish to be tested according to the discrimination results. If the identification result is a female shellfish, the shellfish to be tested is transmitted to the female shellfish area; if the identification result is a male shellfish, the shellfish to be tested is transmitted to the male shellfish area, thereby realizing the automatic sorting of male and female shellfish.

[0011] As a preferred embodiment, the automatic posture correction piece can adjust the size and distance of the correction piece according to different species of shellfish to correct the posture of the shellfish. It can be understood that the shellfish body is stuck in the mold groove of the correction piece by vibration, rolling, etc., and then passes through the baffle with holes through the conveyor belt to reach the next link. The shellfish with incorrect posture will not be stuck in the mold groove but will be pushed back by the baffle, and continue to be stuck in the mold groove through vibration and rolling, thereby achieving posture correction of different species of shellfish.

[0012] Preferably, the X-ray scanning unit includes an X-ray source, a sample bin and a digital imaging plate, wherein the X-ray source and the digital imaging plate are respectively located at the upper and lower parts of the sample bin, and the conveyor belt delivers only one shellfish into the sample bin at a time, and the shell surface of the shellfish is perpendicular to the direction of the X-ray source.

[0013] Preferably, the resolution of the X-ray scanning unit is less than 50 microns.

[0014] Preferably, the control unit is connected to the water conveyor belt via an electrical signal to control the transmission of the water conveyor belt.

[0015] As a preferred method, the male and female shellfish intelligent identification unit performs threshold segmentation on the male and female characteristic images of the shellfish gonads, obtains the pixel features of the male and female shellfish gonads at different developmental stages and constructs a male and female shellfish image database, and then effectively distinguishes the male and female conditions of mussels and the gonad development conditions through machine vision and deep learning algorithms. It can be understood that the whole process mainly includes the following steps: First, prepare training and test data sets. By collecting and annotating data by yourself, establish a sufficient number of image sets containing two categories of male and female so that the model can learn the characteristics of each category. Second, preprocess the image. Before the image is input into the model, the generalization ability and training efficiency of the model are improved by adjusting the size, normalization, enhancement (such as rotation, scaling, flipping, etc.) of the image, and possible operations such as cropping and denoising. Third, according to the needs in the specific operation, flexibly adjust the architecture of the model, such as increasing or decreasing convolutional layers, pooling layers, fully connected layers, etc., and select the VGG commonly used model as the CNN model for image recognition. Fourth, use the prepared data set to train the model. By setting parameters such as learning rate, batch size, number of iterations, and selecting a suitable optimization algorithm (such as SGD, Adam) to update the weights of the model, the model successfully learns the mapping relationship between the features of different gonad images of males and females and the male and female labels. After the training is completed, an independent test set is used to evaluate the performance of the model. Evaluation indicators usually include accuracy, recall rate, F1 score, etc., and the model is adjusted or the data preprocessing is improved according to the results, and further optimization is performed until the algorithm effectively distinguishes the male and female conditions and gonad development of mussels. It is understandable that the training method is a method well known to those skilled in the art, and the male and female conditions and gonad development of mussels can be effectively distinguished based on the above steps and the evaluation indicators established.

[0016] Preferably, the shellfish at different developmental stages are shellfish whose gonads have developed to the differentiation stage and the maturity stage.

[0017] The present invention also provides an automatic sorting method for realizing non-destructive detection of shellfish sex using the automatic sorting system described in any one of the above technical solutions, comprising the following steps:

[0018] The shellfish positioning correction unit is used to correct the posture of the shellfish to be tested one by one sent by the water conveyor belt, and then sent to the X-ray scanning unit via the water conveyor belt;

[0019] The gonads of the shellfish to be tested are scanned by using an X-ray scanning unit to obtain the male and female characteristic images of the gonads of the shellfish;

[0020] The established intelligent shellfish sex identification unit is used to identify the acquired male and female characteristic images of the shellfish gonads to obtain the shellfish sex identification result;

[0021] The automatic sorting unit uses the above-mentioned male-female discrimination results to automatically sort the current shellfish to be tested. If the identification result is a female shellfish, the current shellfish to be tested is transferred to the female shellfish area; if the identification result is a male shellfish, the current shellfish to be tested is transferred to the male shellfish area, thereby realizing automatic sorting of male and female shellfish for non-destructive testing.

[0022] Preferably, the shellfish is a bivalve shellfish.

[0023] Preferably, the accuracy of the method reaches over 95%.

[0024] Compared with the prior art, the advantages and positive effects of the present invention are:

[0025] The automatic sorting system for shellfish sex detection based on X-ray non-destructive detection provided by the present invention can separate male and female individuals for temporary breeding by physical means on the basis of accurately determining the sex of each individual shellfish, and can effectively reduce the problem of low hatching rate caused by excessive sperm discharge caused by too many male individuals, and at the same time, can appropriately reduce the number of male individuals cultivated, effectively improve the hatching rate of shellfish seed production and reduce production costs, improve the comprehensive benefits of shellfish seed production, and provide technical support for shellfish breeding and seed production. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A schematic diagram of an automatic sorting system for non-destructive detection of shellfish sex provided by an embodiment of the present invention;

[0027] Figure 2 The X-ray gonad male and female characteristic images of mussels provided in the embodiment of the present invention;

[0028] Figure 3 This is an X-ray image of the male and female characteristics of the oyster gonads provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] Example 1

[0031] Select mussels as the test object and carry out automatic sorting of mussel sex non-destructive detection. The overall process is as follows: Figure 1 shown.

[0032] The first step is to build an intelligent system for identifying male and female mussels. In winter, 1,000 healthy blue mussels of similar size are collected and placed in the shellfish breeding workshop to start gonad maturation. When the gonads develop to the differentiation and maturity stages, the database of male and female shellfish sex image features is built. The mussel gonads are scanned using an X-ray scanning unit with a resolution of less than 50 microns (for example, the XDR-AZ1600 X-ray machine from Shanghai Xianwei Optoelectronics). The appropriate mussel posture and depth of field are manually adjusted in the early stage to ensure that the X-rays just pass through the shells and obtain the X-ray gonad male and female feature images of the mussels, such as Figure 2 shown.

[0033] Combined with the different characteristics of male and female gonadal tissues at different gonadal development stages of mussels, the convolutional neural network is used to identify the male and female characteristic images of the gonads. First, by collecting and annotating data by yourself, a sufficient number of image sets containing male and female categories are established so that the model can learn the characteristics of each category. Then the image is preprocessed. Before the image is input into the model, the generalization ability and training efficiency of the model are improved by adjusting the image size, normalization, enhancement (such as rotation, scaling, flipping, etc.) and possible cropping and denoising. Next, according to the needs of specific operations, the architecture of the model is flexibly adjusted, such as adding or reducing convolutional layers, pooling layers, fully connected layers, etc., and the VGG commonly used model is selected as the CNN model for image recognition. Then, the model is trained using the prepared data set. By setting parameters such as learning rate, batch size, number of iterations, and selecting the Adam optimization algorithm to update the model weights, the model successfully learns the mapping relationship between the characteristics of different gonadal images of male and female and the male and female labels. After training, an independent test set is used to evaluate the performance of the model. Evaluation indicators usually include accuracy, recall rate, F1 score, etc. Model adjustments or data preprocessing are performed based on the results, and further optimization is carried out until the algorithm can effectively distinguish the gender and gonad development of mussels, and finally establish an intelligent system for identifying male and female mussels.

[0034] The second step is the intelligent sex identification and sorting of mussels. 500 mussels with mature gonads were selected for sex identification and sorting. After the mussels to be tested were cut off from the byssus and washed clean, they entered the shellfish positioning and correction unit through the assembly line conveyor belt. The automatic posture correction part of the positioning and correction unit corrected the posture according to the characteristics of the mussel body. The posture of the mussel was corrected by adjusting the size and distance of the correction part, and then transported to the sample bin of the X-ray scanning unit through the assembly line to ensure that the direction of the X-ray is perpendicular to the shell surface of the shellfish, and the ray source and digital imaging plate are located above and below the sample bin respectively. The conveyor belt sends one shellfish into the sample bin each time, and the shell surface is perpendicular to the ray direction. The posture is consistent with the construction of the male and female mussel image database, and the control unit automatically takes pictures and stores them. The control unit controls the X-ray machine emission through a control box and a computer equipped with a dedicated control system and image intelligent analysis software (Shanghai Xianwei Optoelectronics' XDR-AZ1600 X-ray machine comes with analysis software), and then automatically performs preliminary processing on the digital X-ray image. After the male and female intelligent identification system identifies and analyzes the captured image, the male and female signal of the current shellfish is sent to the control unit, and the control unit sends it to the automatic sorting system based on the male and female information. When the automatic sorting system receives the identification result, if the identification result is a female shellfish, it controls the conveyor belt of the assembly line to transport the current shellfish to the female shellfish area; if the identification result is a male shellfish, it controls the conveyor belt of the assembly line to transport the current shellfish to the male shellfish area, thereby finally realizing the automatic sorting of male and female mussels.

[0035] The third step is sex identification and sorting effect detection. The sex of the sorted male and female mussels is identified through dissection and induced spawning to verify the sex identification and sorting effect of the system, with a detection accuracy of more than 95%.

[0036] Example 2

[0037] Oysters were selected as the test objects, and automatic sorting for non-destructive sex detection of oysters was carried out.

[0038] The first step is to build an intelligent male and female oyster identification system. In winter, 1,000 healthy oysters of similar size are collected and placed in the shellfish breeding workshop to start gonad maturation. When the gonads develop to the differentiation and maturity stages, the oyster male and female gender image feature database is built. An X-ray scanning unit with a resolution of less than 50 microns (for example, the XDR-AZ1600 X-ray machine of Shanghai Xianwei Optoelectronics) is used to scan the mussel gonads. The appropriate oyster posture and depth of field are manually adjusted in the early stage to ensure that the X-rays just pass through the shells and obtain the oyster X-ray gonad male and female feature images, such as Figure 3 shown.

[0039] Combined with the different characteristics of male and female gonadal tissues at different gonadal development stages of oysters, the convolutional neural network is used to identify the male and female characteristic images of the gonads. Specifically, by collecting and annotating data by yourself, a sufficient number of image sets containing male and female categories are established so that the model can learn the characteristics of each category. Then the image is preprocessed. Before the image is input into the model, the generalization ability and training efficiency of the model are improved by adjusting the image size, normalization, enhancement (such as rotation, scaling, flipping, etc.) and possible cropping and denoising. Next, according to the needs of specific operations, the architecture of the model is flexibly adjusted, such as adding or reducing convolutional layers, pooling layers, fully connected layers, etc., and the VGG commonly used model is selected as the CNN model for image recognition. Then the model is trained using the prepared data set. By setting parameters such as learning rate, batch size, number of iterations, and selecting the Adam optimization algorithm to update the weights of the model, the model successfully learns the mapping relationship between the characteristics of different gonadal images of male and female and the male and female labels. After training, an independent test set is used to evaluate the performance of the model. Evaluation indicators usually include accuracy, recall rate, F1 score, etc. The model is adjusted or data preprocessing is improved based on the results, and further optimization is carried out until the algorithm can effectively distinguish the gender and gonad development of oysters, and finally establish an intelligent male and female mussel identification system.

[0040] The second step is the intelligent sex identification and sorting of oysters. 500 single oysters with mature gonads were selected for sex identification and sorting. After the oysters to be tested were washed clean, they entered the shellfish positioning and correction unit through the assembly line conveyor belt. The automatic posture correction part of the positioning and correction unit corrected the posture according to the characteristics of the oyster shell body. The posture of the oyster was corrected by adjusting the size and distance of the correction part. Then, it was transported to the sample bin of the X-ray scanning unit through the assembly line to ensure that the direction of the X-ray was perpendicular to the shell surface of the shellfish, and the ray source and digital imaging plate were located above and below the sample bin respectively. The conveyor belt sent one oyster into the sample bin each time, and the shell surface was perpendicular to the ray direction. The posture was consistent with the construction of the male and female oyster image database, and the control unit automatically took pictures and stored them. The control unit controls the X-ray machine emission through a control box and a computer equipped with a dedicated control system and image intelligent analysis software (Shanghai Xianwei Optoelectronics' XDR-AZ1600 X-ray machine comes with analysis software), and then automatically performs preliminary processing on the digital X-ray image. After the male and female intelligent identification system identifies and analyzes the captured image, the male and female signal of the current shellfish is sent to the control unit, and the control unit sends it to the automatic sorting system based on the male and female information. When the automatic sorting system receives the identification result, if the identification result is a female shellfish, it controls the conveyor belt of the assembly line to transport the current shellfish to the female shellfish area; if the identification result is a male shellfish, it controls the conveyor belt of the assembly line to transport the current shellfish to the male shellfish area, thereby finally realizing the automatic sorting of male and female mussels.

[0041] The third step is sex identification and sorting effect detection. The sex of the sorted male and female oysters is identified through dissection and induced spawning to verify the sex identification and sorting effect of the system, with a detection accuracy of more than 95%.

Claims

1. An automatic sorting system for non-destructive detection of shellfish sex, characterized in that: Includes the following parts: The shellfish positioning and correction unit comprises a water conveyor belt and an automatic posture correction part, wherein the automatic posture correction part is used to correct the posture of the shellfish to be tested that are fed one by one by the water conveyor belt; An X-ray scanning unit is connected to the shellfish positioning and correction unit through a water conveyor belt, and is used to scan the gonads of the shellfish to be tested to obtain the male and female characteristic images of the shellfish gonads; A control unit is connected to the X-ray scanning unit through an electrical signal, and is used to control the X-ray scanning unit to perform real-time imaging and store the male and female characteristic images of the gonads of shellfish in real time; The shellfish sex intelligent identification unit is connected to the control unit through an electrical signal, and is used to identify the sex characteristic images of the shellfish gonads stored in the control unit, and feed back the shellfish sex identification results to the automatic sorting unit; The automatic sorting unit is connected to the control unit through electrical signals, and is used to receive the male and female discrimination results of the shellfish, and automatically sort the shellfish to be tested according to the discrimination results. If the identification result is a female shellfish, the shellfish to be tested is transmitted to the female shellfish area; if the identification result is a male shellfish, the shellfish to be tested is transmitted to the male shellfish area, thereby realizing the automatic sorting of male and female shellfish.

2. The automatic sorting system according to claim 1, characterized in that: The automatic posture correcting piece can adjust the size and distance of the correcting piece according to different species of shellfish to correct the posture of the shell.

3. The automatic sorting system according to claim 1, characterized in that: The X-ray scanning unit includes an X-ray source, a sample chamber and a digital imaging plate, wherein the X-ray source and the digital imaging plate are respectively located at the upper and lower parts of the sample chamber, and the water conveyor belt only delivers one shellfish into the sample chamber at a time, and the shell surface of the shellfish is perpendicular to the direction of the X-ray source.

4. The automatic sorting system according to claim 3, characterized in that: The resolution of the X-ray scanning unit is less than 50 microns.

5. The automatic sorting system according to claim 1, characterized in that: The control unit is connected to the water conveyor belt through an electrical signal and is used to control the transmission of the water conveyor belt.

6. The automatic sorting system according to claim 1, characterized in that: The intelligent shellfish male and female identification unit performs threshold segmentation on the male and female characteristic images of the shellfish gonads, obtains the pixel features of the gonads of male and female shellfish at different developmental stages and constructs a shellfish male and female image database, and then uses machine vision and deep learning algorithms to effectively distinguish the gender and gonad development of mussels.

7. The automatic sorting system according to claim 6, characterized in that: Shellfish at different developmental stages include those whose gonads develop to the differentiation stage and the maturity stage.

8. An automatic sorting method for non-destructive detection of shellfish sex using the automatic sorting system according to any one of claims 1 to 7, characterized in that: The following steps are involved: The shellfish positioning correction unit is used to correct the posture of the shellfish to be tested one by one sent by the water conveyor belt, and then sent to the X-ray scanning unit via the water conveyor belt; The gonads of the shellfish to be tested are scanned by using an X-ray scanning unit to obtain the male and female characteristic images of the gonads of the shellfish; The established intelligent shellfish sex identification unit is used to identify the acquired male and female characteristic images of the shellfish gonads to obtain the shellfish sex identification result; The automatic sorting unit uses the above-mentioned male-female discrimination results to automatically sort the current shellfish to be tested. If the identification result is a female shellfish, the current shellfish to be tested is transferred to the female shellfish area; if the identification result is a male shellfish, the current shellfish to be tested is transferred to the male shellfish area, thereby realizing automatic sorting of male and female shellfish for non-destructive testing.

9. The automatic sorting method according to claim 8, characterized in that: The shellfish is a bivalve shellfish.

10. The automatic sorting method according to claim 8 or 9, characterized in that: The accuracy of the method reaches over 95%.

Images