Live fish catching counting method and system based on image processing

Abstract

The application discloses a kind of live fish catching counting method and system based on image processing, for real-time obtaining live fish video frame image set in culture net cage, and accurately count the species and quantity of cultured fish.The method comprises the following steps: real-time obtaining live fish video frame image set and the name of the species of cultured fish;Construct fish feature database;Extract fish feature information in each frame image;Compare the extracted features with the database, determine the species of fish and mark;Based on trajectory tracking algorithm, construct fish motion trajectory model;Analysis and statistics are carried out on the fish motion trajectory, and the quantity of each kind of fish is obtained.The application realizes automatic counting by using image processing technology, and provides more information through trajectory analysis.The method can improve the accuracy and efficiency of counting, and has important significance for aquaculture management and environmental monitoring.

Description

Technical Field

[0001] This invention relates to the field of fish counting technology, and in particular to a method and system for counting live fish caught based on image processing. Background Technology

[0002] In aquaculture, fish counting is a crucial task, significant for fish management, evaluation of aquaculture efficiency, and environmental monitoring. Traditional fish counting methods typically rely on visual inspection or manual counting, which is time-consuming and susceptible to subjective factors, leading to errors. With the development of computer vision and image processing technologies, automated fish counting using computers has become possible. Currently, some image processing-based fish counting methods exist, some of which identify and count fish based on morphological, color, or texture features. However, these methods are often affected by factors such as lighting conditions, changing viewing angles, and fish movement, resulting in inaccurate and unstable counting results. Therefore, there is a need for an image processing-based live fish harvesting and counting method and system that can accurately count the number of various fish species in aquaculture cages by analyzing fish movement trajectories after acquiring real-time video frame images of live fish in the cages. Summary of the Invention

[0003] To address at least one of the aforementioned technical problems, this invention proposes a live fish capture counting method and system based on image processing.

[0004] The first aspect of this invention provides a live fish capture counting method based on image processing, comprising:

[0005] Real-time acquisition of live fish video frame images and names of farmed fish species in aquaculture cages;

[0006] The first fish feature is obtained based on the name of the farmed fish species, and the first fish feature is used to construct a fish feature database.

[0007] Feature extraction is performed on each frame of the live fish video frame image set to extract fish feature information and obtain the second fish feature;

[0008] The second fish feature is compared with the first fish feature in the fish information database to obtain the fish species, and each fish species is marked in each video frame image;

[0009] A fish motion trajectory model is constructed based on a trajectory tracking algorithm. Each labeled video frame image is imported into the fish trajectory model to obtain the fish motion trajectory.

[0010] By analyzing and statistically analyzing the movement trajectories of fish, the quantity of each type of fish before live fish are harvested can be obtained.

[0011] In this solution, the real-time acquisition of live fish video frame image sets and the names of farmed fish species in aquaculture cages specifically includes:

[0012] Underwater monitoring equipment is installed at the inner edge of the aquaculture cage to acquire video frame images of live fish in the aquaculture cage within a preset time period;

[0013] Obtain the names of the farmed fish in the cages based on the cage aquaculture information.

[0014] In this solution, obtaining the first fish characteristic based on the name of the farmed fish species and constructing the first fish characteristic into a fish characteristic database specifically involves:

[0015] Based on the name of the farmed fish, the corresponding fish characteristic information is retrieved from the Internet to obtain the first fish characteristic, which includes the fish color, size, outline and texture characteristics;

[0016] Construct a fish feature database and import the first fish feature into the fish feature database.

[0017] In this scheme, the step of extracting features from each frame of the live fish video image set to obtain fish feature information and thus the second fish feature is as follows:

[0018] Each frame of the live fish video frame image set is preprocessed, including image denoising and image resizing;

[0019] A deep learning-based target detection algorithm is used to detect fish targets in each preprocessed frame of video image and to mark the position and bounding box of the fish.

[0020] Based on computer vision algorithms, features of fish are extracted from their positions and bounding boxes to obtain second fish features, which include fish color, size, outline, and texture features.

[0021] In this scheme, the step of comparing the second fish feature with the first fish feature in the fish information database to obtain the fish species in each video frame image, and marking each fish species in each video frame image, specifically involves:

[0022] The first and second fish features are imported into a computer for dimensionality reduction processing to obtain the first and second fish feature descriptors.

[0023] The first fish feature descriptor and the second fish feature descriptor are compared to identify the fish species in each video frame image, and the identification result is obtained.

[0024] Based on the recognition results, different types of fish are marked with different colors in each video frame.

[0025] In this scheme, the fish motion trajectory model is constructed based on the trajectory tracking algorithm. Each labeled video frame is imported into the fish trajectory model to obtain the fish motion trajectory. Specifically:

[0026] Constructing a fish movement trajectory model based on trajectory tracking algorithm;

[0027] The continuous video frame images are imported into the fish motion trajectory model, and the center point of the fish in each video frame image is marked to obtain the fish center point frame image;

[0028] By stitching together consecutive fish center point frames and continuously analyzing the positional changes of the center point in adjacent fish center point frames, the movement trajectories of fish with different colors can be obtained.

[0029] In this scheme, the analysis and statistical analysis of fish movement trajectories to obtain the quantity of each type of fish before live fish are harvested is specifically as follows:

[0030] The fish's movement trajectory is preprocessed, including trajectory smoothing and invalid trajectory removal.

[0031] The pre-processed fish movement trajectories are classified by color and divided into different trajectory groups;

[0032] The number of each type of fish before the live fish were caught was obtained by statistically analyzing the trajectory groups.

[0033] A second aspect of the present invention also provides a live fish capture counting system based on image processing. The system includes a memory and a processor. The memory includes a live fish capture counting method program based on image processing. When the processor executes the live fish capture counting method program based on image processing, it performs the following steps:

[0034] Real-time acquisition of live fish video frame images and names of farmed fish species in aquaculture cages;

[0035] The first fish feature is obtained based on the name of the farmed fish species, and the first fish feature is used to construct a fish feature database.

[0036] Feature extraction is performed on each frame of the live fish video frame image set to extract fish feature information and obtain the second fish feature;

[0037] The second fish feature is compared with the first fish feature in the fish information database to obtain the fish species, and each fish species is marked in each video frame image;

[0038] A fish motion trajectory model is constructed based on a trajectory tracking algorithm. Each labeled video frame image is imported into the fish trajectory model to obtain the fish motion trajectory.

[0039] By analyzing and statistically analyzing the movement trajectories of fish, the quantity of each type of fish before live fish are harvested can be obtained.

[0040] In this scheme, the step of extracting features from each frame of the live fish video image set to obtain fish feature information and thus the second fish feature is as follows:

[0041] Each frame of the live fish video frame image set is preprocessed, including image denoising and image resizing;

[0042] A deep learning-based target detection algorithm is used to detect fish targets in each preprocessed frame of video image and to mark the position and bounding box of the fish.

[0043] Based on computer vision algorithms, features of fish are extracted from their positions and bounding boxes to obtain second fish features, which include fish color, size, outline, and texture features.

[0044] In this scheme, the analysis and statistical analysis of fish movement trajectories to obtain the quantity of each type of fish before live fish are harvested is specifically as follows:

[0045] The fish's movement trajectory is preprocessed, including trajectory smoothing and invalid trajectory removal.

[0046] The pre-processed fish movement trajectories are classified by color and divided into different trajectory groups;

[0047] The number of each type of fish before the live fish were caught was obtained by statistically analyzing the trajectory groups.

[0048] This invention discloses a live fish harvesting and counting method and system based on image processing. It is used to acquire a set of live fish video frames in aquaculture cages in real time and accurately count the species and quantity of farmed fish. The method includes the following steps: acquiring a set of live fish video frames and the names of farmed fish species in real time; constructing a fish feature database; extracting fish feature information from each frame; comparing the extracted features with the database to determine and label the fish species; constructing a fish movement trajectory model based on a trajectory tracking algorithm; and analyzing and statistically analyzing the fish movement trajectories to obtain the quantity of each fish species. This invention utilizes image processing technology to achieve automated counting and provides more information through trajectory analysis. This method can improve the accuracy and efficiency of counting and is of great significance for aquaculture management and environmental monitoring. Attached Figure Description

[0049] Figure 1A flowchart of a live fish capture and counting method based on image processing according to the present invention is shown;

[0050] Figure 2 The flowchart illustrating the fish movement trajectory derived by this invention is shown.

[0051] Figure 3 This invention provides a flowchart illustrating the quantity of each type of fish before live fish are harvested.

[0052] Figure 4 A block diagram of a live fish capture and counting system based on image processing according to the present invention is shown. Detailed Implementation

[0053] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0054] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0055] Figure 1 A flowchart of a live fish capture and counting method based on image processing according to the present invention is shown.

[0056] like Figure 1 As shown, the first aspect of the present invention provides a live fish capture counting method based on image processing, comprising:

[0057] S102, real-time acquisition of live fish video frame image set and the name of farmed fish species in aquaculture cages;

[0058] S104, Obtain the first fish feature based on the name of the farmed fish species, and construct the first fish feature into a fish feature database;

[0059] S106, extract features from each frame of the live fish video frame image set to extract fish feature information and obtain the second fish feature;

[0060] S108, compare the second fish feature with the first fish feature in the fish information database to obtain the fish species, and mark each fish species in each video frame image;

[0061] S110, a fish motion trajectory model is constructed based on the trajectory tracking algorithm. Each marked video frame image is imported into the fish trajectory model to obtain the fish motion trajectory.

[0062] S112, analyze and statistically analyze the movement trajectory of fish to obtain the quantity of each type of fish before live fish are caught.

[0063] According to an embodiment of the present invention, the real-time acquisition of the set of live fish video frames and the names of farmed fish species in the aquaculture cages specifically includes:

[0064] Underwater monitoring equipment is installed at the inner edge of the aquaculture cage to acquire video frame images of live fish in the aquaculture cage within a preset time period;

[0065] Obtain the names of the farmed fish in the cages based on the cage aquaculture information.

[0066] According to an embodiment of the present invention, the step of obtaining a first fish feature based on the name of the farmed fish species and constructing the first fish feature into a fish feature database specifically includes:

[0067] Based on the name of the farmed fish, the corresponding fish characteristic information is retrieved from the Internet to obtain the first fish characteristic, which includes the fish color, size, outline and texture characteristics;

[0068] Construct a fish feature database and import the first fish feature into the fish feature database.

[0069] According to an embodiment of the present invention, the step of extracting features from each frame of the live fish video frame image set to extract fish feature information and obtain a second fish feature specifically involves:

[0070] Each frame of the live fish video frame image set is preprocessed, including image denoising and image resizing;

[0071] A deep learning-based target detection algorithm is used to detect fish targets in each preprocessed frame of video image and to mark the position and bounding box of the fish.

[0072] Based on computer vision algorithms, features of fish are extracted from their positions and bounding boxes to obtain second fish features, which include fish color, size, outline, and texture features.

[0073] It should be noted that the video frame image set is obtained by extracting each frame of video image after shooting live fish; the preprocessing can eliminate noise interference in the image and ensure that the subsequent target detection algorithm can accurately identify fish targets.

[0074] According to an embodiment of the present invention, the step of comparing the second fish feature with the first fish feature in the fish information database to obtain the fish species in each video frame image, and marking each fish species in each video frame image, specifically involves:

[0075] The first and second fish features are imported into a computer for dimensionality reduction processing to obtain the first and second fish feature descriptors.

[0076] The first fish feature descriptor and the second fish feature descriptor are compared to identify the fish species in each video frame image, and the identification result is obtained.

[0077] Based on the recognition results, different types of fish are marked with different colors in each video frame.

[0078] It should be noted that the dimensionality reduction process refers to reducing the dimensionality of the first and second fish features to numerical representations, thereby reducing the dimension of the feature vectors and improving the efficiency of subsequent comparisons. The comparison of the first and second fish feature descriptors uses a similarity measurement method. By comparing the similarity between the two feature descriptors, the fish species in each video frame can be identified.

[0079] Figure 2 A flowchart illustrating the fish movement trajectory derived by this invention is shown.

[0080] According to an embodiment of the present invention, the fish motion trajectory model is constructed based on a trajectory tracking algorithm, and the marked video frame images are imported into the fish trajectory model to obtain the fish motion trajectory, specifically as follows:

[0081] S202, Constructing a fish movement trajectory model based on a trajectory tracking algorithm;

[0082] S204, import the continuous video frame images into the fish motion trajectory model, mark the fish center point in each video frame image, and obtain the fish center point frame image;

[0083] S206, stitch together consecutive fish center point frame images, and continuously analyze the positional changes of the center point in adjacent fish center point frame images to obtain the movement trajectories of fish with different colors.

[0084] It should be noted that the trajectory tracking algorithm includes a Kalman filter; the center point marking is achieved by using a machine learning algorithm to identify fish with rectangular bounding boxes, and taking the center point of the rectangle as the center point of the fish; the continuous fish center point frame images are stitched together to form a complete fish motion trajectory sequence, and then, by analyzing the position change of the center point position in adjacent fish center point frame images, fish motion trajectories with different colors can be obtained.

[0085] Figure 3 The flowchart illustrating the process of obtaining the quantity of each type of fish before harvesting live fish according to the present invention is shown.

[0086] According to an embodiment of the present invention, the step of analyzing and statistically analyzing the movement trajectories of fish to obtain the quantity of each type of fish before live fish are harvested specifically involves:

[0087] S302, preprocess the fish's movement trajectory, the preprocessing including trajectory smoothing and invalid trajectory removal;

[0088] S304 classifies the pre-processed fish movement trajectory by color into different trajectory groups;

[0089] S306, The trajectory group is statistically analyzed to obtain the quantity of each type of fish before the live fish are caught.

[0090] It should be noted that the preprocessing can remove invalid and incomplete trajectories, ensuring the accuracy and reliability of the analysis process; the fish movement trajectories are classified by color, with one color representing one type of fish; the trajectory group is composed of the movement trajectories of the same species; the number of fish is obtained by calculating the number of trajectories, with one trajectory representing one fish.

[0091] According to an embodiment of the present invention, it further includes:

[0092] An underwater robot was used to take 360-degree photos of the aquaculture cages, obtaining a complete image of the surface of the aquaculture cages.

[0093] Obtain images of the surface of non-destructive aquaculture cages;

[0094] Import the complete image and the undamaged image of the aquaculture cage surface into the computer to draw the net wires of the aquaculture cage and the undamaged aquaculture cage, and obtain the drawn image of the aquaculture cage and the drawn image of the undamaged aquaculture cage.

[0095] Compare the drawn images of aquaculture cages with the drawn images of undamaged aquaculture cages to determine whether the aquaculture cages are damaged.

[0096] If the aquaculture cages are damaged, repair them.

[0097] It should be noted that, in this embodiment of the invention, the image of the aquaculture cage is processed to determine whether the aquaculture cage is damaged. If damage is found, the aquaculture cage is repaired to avoid inaccurate counting of live fish.

[0098] According to an embodiment of the present invention, it further includes:

[0099] In each group of trajectories, a preset percentage of fish movement trajectories are extracted, imported into the computer, and calculated to obtain the length of the fish movement trajectory.

[0100] Based on the recording duration of video frame images, the duration of the fish's movement trajectory can be obtained;

[0101] The fish's speed is calculated based on the length of its trajectory and the duration of its movement.

[0102] The average speed of each group of fish is calculated by averaging the calculated speeds of each fish species.

[0103] The activity of each type of fish is evaluated based on the average movement speed to obtain a fish activity index;

[0104] Based on the fish activity index, the duration and frame rate of acquiring video frames of each type of fish are adjusted. If the fish activity index is less than a preset value, the duration and frame rate of acquiring video frames of fish are reduced.

[0105] It should be noted that, in this embodiment of the invention, by analyzing the movement activity of the fish school, the duration and frame rate of the video frames acquired for each type of fish are adjusted. The higher the movement activity of the fish school, the longer the duration of the acquired video frames and the higher the frame rate; the lower the movement activity of the fish school, the shorter the duration of the acquired video frames and the lower the frame rate. This helps to reduce the amount of image analysis, reduce the amount of data for fish school analysis, and improve analysis efficiency and accuracy. The preset percentage refers to the percentage calculated by the management personnel based on the number of fish.

[0106] According to an embodiment of the present invention, it further includes:

[0107] Within a preset time period, acquire real-time images of live fish entering the transfer pipe after they have been caught;

[0108] Fish images are obtained by performing fish identification on the real-time images based on the target detection algorithm.

[0109] Based on the fish images, the number of fish was counted to obtain the number of fish that entered the transfer pipe;

[0110] The number of fish is calculated in conjunction with a preset time to obtain the fish transfer flow rate per second.

[0111] The total transfer time of the fish is obtained, and the total number of fish after the live fish are caught is calculated based on the total time and the transfer flow rate.

[0112] The total number of fish species before live fish are caught is obtained by counting the number of each species before live fish are caught.

[0113] The deviation value is calculated based on the total number of fish before and after the live fish are caught.

[0114] If the deviation value is outside the preset range, the total number of fish will be corrected.

[0115] It should be noted that the preset time refers to one minute; the correction operation is to calculate the average of the total number of fish before and after the live fish are caught to obtain the corrected number of fish; in this embodiment of the invention, the number of fish after the live fish are caught can be counted to further determine the number of fish and make the number of fish more accurate; the target detection algorithm includes the YOLO algorithm.

[0116] Figure 4 A block diagram of a live fish capture counting system based on image processing is shown.

[0117] A second aspect of the present invention also provides a live fish capture counting system 4 based on image processing. The system includes a memory 41 and a processor 42. The memory includes a live fish capture counting method program based on image processing. When the processor executes the live fish capture counting method program based on image processing, it performs the following steps:

[0118] Real-time acquisition of live fish video frame images and names of farmed fish species in aquaculture cages;

[0119] The first fish feature is obtained based on the name of the farmed fish species, and the first fish feature is used to construct a fish feature database.

[0120] Feature extraction is performed on each frame of the live fish video frame image set to extract fish feature information and obtain the second fish feature;

[0121] The second fish feature is compared with the first fish feature in the fish information database to obtain the fish species, and each fish species is marked in each video frame image;

[0122] A fish motion trajectory model is constructed based on a trajectory tracking algorithm. Each labeled video frame image is imported into the fish trajectory model to obtain the fish motion trajectory.

[0123] By analyzing and statistically analyzing the movement trajectories of fish, the quantity of each type of fish before live fish are harvested can be obtained.

[0124] According to an embodiment of the present invention, the real-time acquisition of the set of live fish video frames and the names of farmed fish species in the aquaculture cages specifically includes:

[0125] Underwater monitoring equipment is installed at the inner edge of the aquaculture cage to acquire video frame images of live fish in the aquaculture cage within a preset time period;

[0126] Obtain the names of the farmed fish in the cages based on the cage aquaculture information.

[0127] According to an embodiment of the present invention, the step of obtaining a first fish feature based on the name of the farmed fish species and constructing the first fish feature into a fish feature database specifically includes:

[0128] Based on the name of the farmed fish, the corresponding fish characteristic information is retrieved from the Internet to obtain the first fish characteristic, which includes the fish color, size, outline and texture characteristics;

[0129] Construct a fish feature database and import the first fish feature into the fish feature database.

[0130] According to an embodiment of the present invention, the step of extracting features from each frame of the live fish video frame image set to extract fish feature information and obtain a second fish feature specifically involves:

[0131] Each frame of the live fish video frame image set is preprocessed, including image denoising and image resizing;

[0132] A deep learning-based target detection algorithm is used to detect fish targets in each preprocessed frame of video image and to mark the position and bounding box of the fish.

[0133] Based on computer vision algorithms, features of fish are extracted from their positions and bounding boxes to obtain second fish features, which include fish color, size, outline, and texture features.

[0134] It should be noted that the video frame image set is obtained by extracting each frame of video image after shooting live fish; the preprocessing can eliminate noise interference in the image and ensure that the subsequent target detection algorithm can accurately identify fish targets.

[0135] According to an embodiment of the present invention, the step of comparing the second fish feature with the first fish feature in the fish information database to obtain the fish species in each video frame image, and marking each fish species in each video frame image, specifically involves:

[0136] The first and second fish features are imported into a computer for dimensionality reduction processing to obtain the first and second fish feature descriptors.

[0137] The first fish feature descriptor and the second fish feature descriptor are compared to identify the fish species in each video frame image, and the identification result is obtained.

[0138] Based on the recognition results, different types of fish are marked with different colors in each video frame.

[0139] It should be noted that the dimensionality reduction process refers to reducing the dimensionality of the first and second fish features to numerical representations, thereby reducing the dimension of the feature vectors and improving the efficiency of subsequent comparisons. The comparison of the first and second fish feature descriptors uses a similarity measurement method. By comparing the similarity between the two feature descriptors, the fish species in each video frame can be identified.

[0140] According to an embodiment of the present invention, the fish motion trajectory model is constructed based on a trajectory tracking algorithm, and the marked video frame images are imported into the fish trajectory model to obtain the fish motion trajectory, specifically as follows:

[0141] Constructing a fish movement trajectory model based on trajectory tracking algorithm;

[0142] The continuous video frame images are imported into the fish motion trajectory model, and the center point of the fish in each video frame image is marked to obtain the fish center point frame image;

[0143] By stitching together consecutive fish center point frames and continuously analyzing the positional changes of the center point in adjacent fish center point frames, the movement trajectories of fish with different colors can be obtained.

[0144] It should be noted that the trajectory tracking algorithm includes a Kalman filter; the center point marking is achieved by using a machine learning algorithm to identify fish with rectangular bounding boxes, and taking the center point of the rectangle as the center point of the fish; the continuous fish center point frame images are stitched together to form a complete fish motion trajectory sequence, and then, by analyzing the position change of the center point position in adjacent fish center point frame images, fish motion trajectories with different colors can be obtained.

[0145] According to an embodiment of the present invention, the step of analyzing and statistically analyzing the movement trajectories of fish to obtain the quantity of each type of fish before live fish are harvested specifically involves:

[0146] The fish's movement trajectory is preprocessed, including trajectory smoothing and invalid trajectory removal.

[0147] The pre-processed fish movement trajectories are classified by color and divided into different trajectory groups;

[0148] The number of each type of fish before the live fish were caught was obtained by statistically analyzing the trajectory groups.

[0149] It should be noted that the preprocessing can remove invalid and incomplete trajectories, ensuring the accuracy and reliability of the analysis process; the fish movement trajectories are classified by color, with one color representing one type of fish; the trajectory group is composed of the movement trajectories of the same species; the number of fish is obtained by calculating the number of trajectories, with one trajectory representing one fish.

[0150] According to an embodiment of the present invention, it further includes:

[0151] An underwater robot was used to take 360-degree photos of the aquaculture cages, obtaining a complete image of the surface of the aquaculture cages.

[0152] Obtain images of the surface of non-destructive aquaculture cages;

[0153] Import the complete image and the undamaged image of the aquaculture cage surface into the computer to draw the net wires of the aquaculture cage and the undamaged aquaculture cage, and obtain the drawn image of the aquaculture cage and the drawn image of the undamaged aquaculture cage.

[0154] Compare the drawn images of aquaculture cages with the drawn images of undamaged aquaculture cages to determine whether the aquaculture cages are damaged.

[0155] If the aquaculture cages are damaged, repair them.

[0156] It should be noted that, in this embodiment of the invention, the image of the aquaculture cage is processed to determine whether the aquaculture cage is damaged. If damage is found, the aquaculture cage is repaired to avoid inaccurate counting of live fish.

[0157] According to an embodiment of the present invention, it further includes:

[0158] In each group of trajectories, a preset percentage of fish movement trajectories are extracted, imported into the computer, and calculated to obtain the length of the fish movement trajectory.

[0159] Based on the recording duration of video frame images, the duration of the fish's movement trajectory can be obtained;

[0160] The fish's speed is calculated based on the length of its trajectory and the duration of its movement.

[0161] The average speed of each group of fish is calculated by averaging the calculated speeds of each fish species.

[0162] The activity of each type of fish is evaluated based on the average movement speed to obtain a fish activity index;

[0163] Based on the fish activity index, the duration and frame rate of acquiring video frames of each type of fish are adjusted. If the fish activity index is less than a preset value, the duration and frame rate of acquiring video frames of fish are reduced.

[0164] It should be noted that, in this embodiment of the invention, by analyzing the movement activity of the fish school, the duration and frame rate of the video frames acquired for each type of fish are adjusted. The higher the movement activity of the fish school, the longer the duration of the acquired video frames and the higher the frame rate; the lower the movement activity of the fish school, the shorter the duration of the acquired video frames and the lower the frame rate. This helps to reduce the amount of image analysis, reduce the amount of data for fish school analysis, and improve analysis efficiency and accuracy. The preset percentage refers to the percentage calculated by the management personnel based on the number of fish.

[0165] According to an embodiment of the present invention, it further includes:

[0166] Within a preset time period, acquire real-time images of live fish entering the transfer pipe after they have been caught;

[0167] Fish images are obtained by performing fish identification on the real-time images based on the target detection algorithm.

[0168] Based on the fish images, the number of fish was counted to obtain the number of fish that entered the transfer pipe;

[0169] The number of fish is calculated in conjunction with a preset time to obtain the fish transfer flow rate per second.

[0170] The total transfer time of the fish is obtained, and the total number of fish after the live fish are caught is calculated based on the total time and the transfer flow rate.

[0171] The total number of fish species before live fish are caught is obtained by counting the number of each species before live fish are caught.

[0172] The deviation value is calculated based on the total number of fish before and after the live fish are caught.

[0173] If the deviation value is outside the preset range, the total number of fish will be corrected.

[0174] It should be noted that the preset time refers to one minute; the correction operation is to calculate the average of the total number of fish before and after the live fish are caught to obtain the corrected number of fish; in this embodiment of the invention, the number of fish after the live fish are caught can be counted to further determine the number of fish and make the number of fish more accurate.

[0175] This invention discloses a live fish harvesting and counting method and system based on image processing. It is used to acquire a set of live fish video frames in aquaculture cages in real time and accurately count the species and quantity of farmed fish. The method includes the following steps: acquiring a set of live fish video frames and the names of farmed fish species in real time; constructing a fish feature database; extracting fish feature information from each frame; comparing the extracted features with the database to determine and label the fish species; constructing a fish movement trajectory model based on a trajectory tracking algorithm; and analyzing and statistically analyzing the fish movement trajectories to obtain the quantity of each fish species. This invention utilizes image processing technology to achieve automated counting and provides more information through trajectory analysis. This method can improve the accuracy and efficiency of counting and is of great significance for aquaculture management and environmental monitoring.

[0176] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0177] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0178] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0179] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0180] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

[0181] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for counting live fish caught based on image processing, characterized in that, Includes the following steps: Real-time acquisition of live fish video frame images and names of farmed fish species in aquaculture cages; The first fish feature is obtained based on the name of the farmed fish species, and the first fish feature is used to construct a fish feature database. Feature extraction is performed on each frame of the live fish video frame image set to extract fish feature information and obtain the second fish feature; The second fish feature is compared with the first fish feature in the fish information database to obtain the fish species, and each fish species is marked in each video frame image; A fish motion trajectory model is constructed based on a trajectory tracking algorithm. Each labeled video frame image is imported into the fish trajectory model to obtain the fish motion trajectory. By analyzing and statistically analyzing the movement trajectories of fish, the quantity of each type of fish before the live fish are caught can be obtained; This also includes: extracting a preset percentage of fish movement trajectories from each trajectory group, importing them into a computer for calculation, and obtaining the length of the fish movement trajectory; Based on the recording duration of video frame images, the duration of the fish's movement trajectory can be obtained; The fish's speed is calculated based on the length of its trajectory and the duration of its movement. The average speed of each group of fish is calculated by averaging the calculated speeds of each fish species. The activity of each type of fish is evaluated based on the average movement speed to obtain a fish activity index; Based on the fish activity index, adjust the duration and frame rate of acquiring video frames of each type of fish. If the fish activity index is less than a preset value, reduce the duration and frame rate of acquiring video frames of fish. This also includes: acquiring real-time images of live fish entering the transfer pipe after being caught within a preset time period; Fish images are obtained by performing fish identification on the real-time images based on the target detection algorithm. Based on the fish images, the number of fish was counted to obtain the number of fish that entered the transfer pipe; The number of fish is calculated in conjunction with a preset time to obtain the fish transfer flow rate per second. The total transfer time of the fish is obtained, and the total number of fish after the live fish are caught is calculated based on the total time and the transfer flow rate. The total number of fish species before live fish are caught is obtained by counting the number of each species before live fish are caught. The deviation value is calculated based on the total number of fish before and after the live fish are caught. If the deviation value is outside the preset range, the total number of fish will be corrected.

2. The live fish capture and counting method based on image processing according to claim 1, characterized in that, The real-time acquisition of live fish video frame images and the names of farmed fish species in aquaculture cages specifically includes: Underwater monitoring equipment is installed at the inner edge of the aquaculture cage to acquire video frame images of live fish in the aquaculture cage within a preset time period; Obtain the names of the farmed fish in the cages based on the cage aquaculture information.

3. The live fish capture counting method based on image processing according to claim 1, characterized in that, The step of obtaining the first fish feature based on the name of the farmed fish species and constructing the first fish feature into a fish feature database specifically involves: Based on the name of the farmed fish, the corresponding fish characteristic information is retrieved from the Internet to obtain the first fish characteristic, which includes the fish color, size, outline and texture characteristics; Construct a fish feature database and import the first fish feature into the fish feature database.

4. The live fish capture counting method based on image processing according to claim 1, characterized in that, The step of extracting features from each frame of the live fish video image set to obtain fish feature information and thus the second fish feature is as follows: Each frame of the live fish video frame image set is preprocessed, including image denoising and image resizing; A deep learning-based target detection algorithm is used to detect fish targets in each preprocessed frame of video image and to mark the position and bounding box of the fish. Based on computer vision algorithms, features of fish are extracted from their positions and bounding boxes to obtain second fish features, which include fish color, size, outline, and texture features.

5. The live fish capture counting method based on image processing according to claim 1, characterized in that, The step of comparing the second fish feature with the first fish feature in the fish information database to obtain the fish species in each video frame image, and marking each fish species in each video frame image, specifically involves: The first and second fish features are imported into a computer for dimensionality reduction processing to obtain the first and second fish feature descriptors. The first fish feature descriptor and the second fish feature descriptor are compared to identify the fish species in each video frame image, and the identification result is obtained. Based on the recognition results, different types of fish are marked with different colors in each video frame.

6. The live fish capture counting method based on image processing according to claim 1, characterized in that, The fish motion trajectory model is constructed based on the trajectory tracking algorithm. Each labeled video frame is imported into the fish trajectory model to obtain the fish motion trajectory, specifically as follows: Constructing a fish movement trajectory model based on trajectory tracking algorithm; The continuous video frame images are imported into the fish motion trajectory model, and the center point of the fish in each video frame image is marked to obtain the fish center point frame image; By stitching together consecutive fish center point frames and continuously analyzing the positional changes of the center point in adjacent fish center point frames, the movement trajectories of fish with different colors can be obtained.

7. The live fish capture counting method based on image processing according to claim 1, characterized in that, The analysis and statistics of fish movement trajectories yielded the quantity of each type of fish before the live fish were harvested, specifically: The fish's movement trajectory is preprocessed, including trajectory smoothing and invalid trajectory removal. The pre-processed fish movement trajectories are classified by color and divided into different trajectory groups; The number of each type of fish before the live fish were caught was obtained by statistically analyzing the trajectory groups.

8. A live fish capture counting system based on image processing, characterized in that, The image processing-based live fish capture counting system includes a storage device and a processor. The storage device includes an image processing-based live fish capture counting method program. When the image processing-based live fish capture counting method program is executed by the processor, it performs the following steps: Real-time acquisition of live fish video frame images and names of farmed fish species in aquaculture cages; The first fish feature is obtained based on the name of the farmed fish species, and the first fish feature is used to construct a fish feature database. Feature extraction is performed on each frame of the live fish video frame image set to extract fish feature information and obtain the second fish feature; The second fish feature is compared with the first fish feature in the fish information database to obtain the fish species, and each fish species is marked in each video frame image; A fish motion trajectory model is constructed based on a trajectory tracking algorithm. Each labeled video frame image is imported into the fish trajectory model to obtain the fish motion trajectory. By analyzing and statistically analyzing the movement trajectories of fish, the quantity of each type of fish before the live fish are caught can be obtained; This also includes: extracting a preset percentage of fish movement trajectories from each trajectory group, importing them into a computer for calculation, and obtaining the length of the fish movement trajectory; Based on the recording duration of video frame images, the duration of the fish's movement trajectory can be obtained; The fish's speed is calculated based on the length of its trajectory and the duration of its movement. The average speed of each group of fish is calculated by averaging the calculated speeds of each fish species. The activity of each type of fish is evaluated based on the average movement speed to obtain a fish activity index; Based on the fish activity index, adjust the duration and frame rate of acquiring video frames of each type of fish. If the fish activity index is less than a preset value, reduce the duration and frame rate of acquiring video frames of fish. This also includes: acquiring real-time images of live fish entering the transfer pipe after being caught within a preset time period; Fish images are obtained by performing fish identification on the real-time images based on the target detection algorithm. Based on the fish images, the number of fish was counted to obtain the number of fish that entered the transfer pipe; The number of fish is calculated in conjunction with a preset time to obtain the fish transfer flow rate per second. The total transfer time of the fish is obtained, and the total number of fish after the live fish are caught is calculated based on the total time and the transfer flow rate. The total number of fish species before live fish are caught is obtained by counting the number of each species before live fish are caught. The deviation value is calculated based on the total number of fish before and after the live fish are caught. If the deviation value is outside the preset range, the total number of fish will be corrected.

9. A live fish capture counting system based on image processing according to claim 8, characterized in that, The step of extracting features from each frame of the live fish video image set to obtain fish feature information and thus the second fish feature is as follows: Each frame of the live fish video frame image set is preprocessed, including image denoising and image resizing; A deep learning-based target detection algorithm is used to detect fish targets in each preprocessed frame of video image and to mark the position and bounding box of the fish. Based on computer vision algorithms, features of fish are extracted from their positions and bounding boxes to obtain second fish features, which include fish color, size, outline, and texture features.

10. A live fish capture counting system based on image processing according to claim 8, characterized in that, The analysis and statistics of fish movement trajectories yielded the quantity of each type of fish before the live fish were harvested, specifically: The fish's movement trajectory is preprocessed, including trajectory smoothing and invalid trajectory removal. The pre-processed fish movement trajectories are classified by color and divided into different trajectory groups; The number of each type of fish before the live fish were caught was obtained by statistically analyzing the trajectory groups.

Images