Automatic pond patrol method and system for crayfish infrared image recognition

Abstract

A method and system for automated pond inspection using infrared image recognition for crayfish ponds is described. This method uses an online infrared image acquisition device to capture images of the pond bank and then processes the images using an image processing unit. The image processing unit comprises an image enhancement module, a binarization module, an edge detection module, a target extraction module, and a target counting module, all connected in sequence. The method and system can identify the number of crayfish on the bank in an image, providing convenience for farmers and saving time and effort.

Description

Technical Field

[0001] The present invention relates to the technical field of crayfish breeding, and in particular to an automatic pond patrol method and system for crayfish infrared image recognition. Background Art

[0002] During crayfish farming, deteriorating water quality and lack of oxygen can cause crayfish to come ashore, requiring farmers to regularly monitor their crayfish. Furthermore, large ponds make it difficult for farmers on the other side to observe the crayfish at night, making it difficult to clearly determine whether the crayfish have come ashore. This significantly increases their workload. Traditional pond inspections rely primarily on farmers visually observing and judging the water's conditions based on the number of crayfish that have come ashore. This process is quite inconvenient and laborious. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide an automatic pond patrol method and system for crayfish infrared image recognition, which can identify the number of crayfish on the shore, bring convenience to the breeding personnel, and save time and effort.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] An automatic pond inspection method based on infrared image recognition of crayfish comprises the following steps:

[0006] Step S1: First, an infrared image acquisition device on the UAV carrier platform captures images of the crayfish breeding pond and the shore;

[0007] Step S2: The captured image is then processed by an image processing unit to identify the number of crayfish on the shore in the image;

[0008] Step S3: Finally, the data is remotely transmitted to the interactive platform via the communication device on the drone carrier platform.

[0009] The image processing unit in step S2 includes the following steps:

[0010] Step 1) First, the image enhancement module performs image denoising to filter out the noise information in the image;

[0011] Step 2), then, the binarization module will use the image after the noise reduction module to perform black and white processing to obtain an image with clear feature contours;

[0012] Step 3), then, the edge detection module obtains the edge in the image to form a feature frame diagram;

[0013] Step 4), finally, the target extraction module is used to detect the number of feature contours and output the number obtained by the target extraction module.

[0014] In step 1), after the image enhancement module grayscales the image, the pixel value of each pixel in the image is recorded. Then, the image is divided into several partition windows. Each partition window consists of a square composed of an odd number of pixels. The pixels in each square are arranged and the median is taken. Finally, several partition windows with the median taken are combined to obtain the denoised image.

[0015] In step 2), the binarization module sets the grayscale value of the pixel points on the image after the noise reduction module to 0 or 255 for image binarization, sets the threshold to 158, sets the value of all pixels greater than 158 in the image to 255, and the rest to 0, to obtain a black and white image.

[0016] In step 3), the edge detection module uses Canny edge detection, performs filter convolution calculation on the image, calculates the gradient amplitude and direction of the image to obtain the total of all possible edges of the image, and then performs non-maximum suppression to eliminate non-edge pixels, retaining only candidate edges of thin lines. Then, double threshold screening selects edges that fall within the threshold range. The double thresholds are set to 85 and 140 respectively, completing the edge processing of the image to form a feature box diagram.

[0017] In step 4), the target extraction module is connected to the training module, which is connected to the feature set construction module. The feature set construction module constructs a feature set of crayfish in the image and feeds it into the model trained in the training module to detect target features in the image.

[0018] The feature set construction module includes a crayfish image input module, a shape acquisition module, a polygon construction module and a feature set output module. The crayfish image input module is used to manually input crayfish images, the shape acquisition module is used to obtain the shape features of crayfish, the polygon construction module is used to obtain the polygon number set formed by the crayfish shape features, and the feature set output module is used to output the obtained crayfish feature set.

[0019] The polygon construction module includes an initial contour construction module, which is connected to a polygon approximation module. The polygon approximation module is respectively connected to a convex hull detection module and an angle contour analysis module. The angle contour analysis module and the convex hull detection module are jointly connected to an intersection set calculation module. The initial contour construction module is used to obtain the initial contour of the crayfish and each contour point on the initial contour. The polygon approximation module performs polygon approximation on the initial contour to reduce the number of contour points of the initial contour to obtain a first point set. The convex hull detection module performs convex hull detection on the contour after polygon approximation so that the contour points located on the abdomen of the shrimp are removed from the first point set to form a second point set. The angle contour analysis module performs angle contour analysis on the contour after polygon approximation to form a third point set. The intersection set calculation module is used to obtain the intersection of the second point set and the third point set, and the intersection is the feature set of the shrimp head and shrimp tail.

[0020] An automatic pond patrol system for crayfish infrared image recognition includes an infrared image acquisition device and an image processing unit installed on a drone carrier platform. The infrared image acquisition device and the image processing unit are wirelessly connected to an interactive platform. The image processing unit includes an image enhancement module, a binarization module, an edge detection module, a target extraction module, and a target counting module connected in sequence.

[0021] The image enhancement module is used for image noise reduction to filter out noise information in the image;

[0022] The binarization module is used to perform black and white processing on the image to obtain an image with clear feature contours;

[0023] The edge detection module is used to obtain edge-forming feature frames in an image;

[0024] The target extraction module is used to detect the number of feature contours;

[0025] The target counting module outputs the number obtained by the target extraction module.

[0026] The present invention provides an automatic pond inspection method and system for infrared image recognition of crayfish, which has the following technical effects:

[0027] 1) An infrared image acquisition device is used to capture infrared images of the crayfish breeding pond and the shore. An image processing unit is used to process the captured images and identify the number of crayfish on the shore in the images, bringing convenience to the breeding staff and saving time and effort.

[0028] 2) When processing the image, the image enhancement module grayscales the image and records the pixel value of each pixel in the image. Then, the image is divided into several partition windows. Each partition window is a square composed of an odd number of pixels. The pixels are arranged in the median in each square. Finally, the image after noise reduction is obtained by combining several partition windows after median. The binarization module sets the grayscale value of the pixel on the image after the noise reduction module to 0 or 255 for image binary processing. The threshold is set to 158. The value of all pixels greater than 158 in the image is set to 255, and the rest are 0 to obtain a black and white image. The edge detection module uses Canny edge detection performs filter convolution on the image, calculates the gradient amplitude and direction of the image to obtain the total of all possible edges of the image, then performs non-maximum suppression to exclude non-edge pixels, retaining only candidate edges of thin lines, and then double-threshold screening selects the edges that fall within the threshold range. The double thresholds are set to 85 and 140 respectively, completing the edge processing of the image to form a feature frame diagram. The target extraction module is used to detect the number of feature contours, and the target counting module outputs the number obtained by the target extraction module. After processing the image, the position of the crayfish becomes a frame diagram of the crayfish contour, which is conducive to accurately identifying the crayfish and calculating the number.

[0029] 3) The crayfish image input module is used to manually input the crayfish image, the shape acquisition module is used to obtain the shape features of the crayfish, the polygon construction module is used to obtain the polygon number set formed by the crayfish shape features, the initial contour construction module is used to obtain the initial contour of the crayfish and the contour points on the initial contour, the polygon approximation module performs polygon approximation on the initial contour to reduce the number of contour points of the initial contour to obtain a first point set, the convex hull detection module performs convex hull detection on the contour after polygon approximation so that the contour points located on the abdomen of the shrimp are removed from the first point set to form a second point set, the angle contour analysis module performs angle contour analysis on the contour after polygon approximation to form a third point set, the intersection set calculation module is used to obtain the intersection of the second point set and the third point set, the intersection is the feature set of the shrimp head and shrimp tail, the feature set output module is used to output the obtained feature set of the shrimp head and shrimp tail, which is more accurate, eliminates the influence of the crayfish abdomen pattern on target detection, and improves the accuracy of target number recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The present invention will be further described below with reference to the accompanying drawings and examples:

[0031] Figure 1 This is a schematic diagram of the present invention.

[0032] Figure 2 This is a schematic diagram of the image processing unit in the present invention.

[0033] Figure 3This is a schematic diagram of the feature set construction module in the present invention.

[0034] Figure 4 This is a schematic diagram of the polygonal building module in the present invention. DETAILED DESCRIPTION

[0035] like Figure 1 As shown, an automatic pond patrol system for crayfish infrared image recognition includes an unmanned aerial vehicle (UAV) carrier platform on which an infrared image acquisition device, an image processing unit, and a communication device are installed.

[0036] The infrared image acquisition device is used to capture images of crayfish breeding ponds and the shore.

[0037] The infrared image acquisition device is an infrared thermal imager. Available models: Yoseen F series airborne infrared thermal imager.

[0038] The image processing unit is used to process the captured image and identify the number of crayfish on the shore in the image.

[0039] The communication device is used for remote data transmission between the UAV carrier platform and the interactive platform end through the communication device.

[0040] The communication device is a WIFI module, and the available model is: SKW77

[0041] The image processing unit includes an image enhancement module, a binarization module, an edge detection module, a target extraction module and a target counting module connected in sequence, wherein: the image enhancement module performs image denoising to filter out the noise information in the image, and then the binarization module performs black and white processing on the image after using the denoising module to obtain an image with clear feature contours, and then the edge detection module obtains the edges in the image to form a feature frame diagram, the target extraction module is used to detect the number of feature contours, and the target counting module outputs the number obtained by the target extraction module.

[0042] After the image enhancement module grayscales the image, it records the pixel value of each pixel in the image, and then divides the image into several partition windows. Each partition window consists of a square composed of an odd number of pixels. In each square, the pixels are arranged and the median is taken. Finally, several partition windows with the median taken are combined to obtain the denoised image.

[0043] The binarization module sets the grayscale value of the pixels on the image after the noise reduction module to 0 or 255 for image binary processing, sets the threshold to 158, sets the value of all pixels greater than 158 in the image to 255, and the rest to 0, to obtain a black and white image.

[0044] The edge detection module uses Canny edge detection, performs filter convolution calculation on the image, calculates the gradient amplitude and direction of the image to obtain the total of all possible edges of the image, then performs non-maximum suppression to eliminate non-edge pixels, retaining only candidate edges of thin lines, and then selects edges that fall within the threshold range through dual threshold screening, with the dual thresholds set to 85 and 140 respectively, completing the edge processing of the image to form a feature box diagram.

[0045] The target extraction module is used to detect the number of characteristic contours. The target extraction module is connected to the training module, which is connected to the feature set construction module. The feature set construction module constructs a feature set of crayfish in the image and feeds it into the training module to train a model for detecting target features in the image.

[0046] The initial mathematical model of the training module is manually input, and the input is the DNN model.

[0047] The following is the code for image recognition prediction after obtaining the trained model, based on C++ use, as follows:

[0048]

[0049]

[0050]

[0051] The feature set construction module includes a crayfish image input module, a shape acquisition module, a polygon construction module and a feature set output module. The crayfish image input module is used to manually input crayfish images, the shape acquisition module is used to obtain the shape features of crayfish, the polygon construction module is used to obtain the polygon number set formed by the crayfish shape features, and the feature set output module is used to output the obtained crayfish feature set.

[0052] The polygon construction module includes an initial contour construction module, the initial contour construction module is connected to the polygon approximation module, the polygon approximation module is respectively connected to the convex hull detection module and the angle contour analysis module, and the angle contour analysis module and the convex hull detection module are jointly connected to the intersection set calculation module.

[0053] The initial contour construction module is used to obtain the initial contour of the crayfish and the contour points on the initial contour. The polygon approximation module performs polygon approximation on the initial contour, thereby reducing the number of contour points in the initial contour to obtain a first point set. The convex hull detection module performs convex hull detection on the contour after polygon approximation, removing the contour points located on the shrimp's abdomen from the first point set, forming a second point set. The angle contour analysis module performs angle contour analysis on the contour after polygon approximation to form a third point set. The intersection point calculation module is used to calculate the intersection of the second and third point sets. The intersection is the feature set of the shrimp head and tail.

Claims

1. An automatic pond inspection method using infrared image recognition of crayfish, comprising the following steps: Step S1: First, an infrared image acquisition device on the UAV carrier platform captures images of the crayfish breeding pond and the shore; Step S2: The captured image is then processed by an image processing unit to identify the number of crayfish on the shore in the image; Step S3: Finally, the data is remotely transmitted to the interactive platform via the communication device on the UAV carrier platform; The image processing unit in step S2 includes the following steps: Step 1) First, the image enhancement module performs image denoising to filter out the noise information in the image; Step 2), then, the binarization module will use the image after the noise reduction module to perform black and white processing to obtain an image with clear feature contours; Step 3), then, the edge detection module obtains the edge in the image to form a feature frame diagram; Step 4), finally, the target extraction module is used to detect the number of feature contours and output the number obtained by the target extraction module; In step 4), the target extraction module is connected to the training module, and the training module is connected to the feature set construction module; the feature set construction module constructs a feature set of crayfish in the image and feeds it into the model trained in the training module to detect target features in the image; The feature set construction module includes a crayfish image input module, a shape acquisition module, a polygon construction module and a feature set output module. The crayfish image input module is used to manually input a crayfish image, the shape acquisition module is used to obtain the shape features of the crayfish, the polygon construction module is used to obtain a polygon set formed by the crayfish shape features, and the feature set output module is used to output the obtained crayfish feature set. The polygon construction module includes an initial contour construction module, which is connected to a polygon approximation module. The polygon approximation module is respectively connected to a convex hull detection module and an angle contour analysis module. The angle contour analysis module and the convex hull detection module are jointly connected to an intersection set calculation module. The initial contour construction module is used to obtain the initial contour of the crayfish and each contour point on the initial contour. The polygon approximation module performs polygon approximation on the initial contour to reduce the number of contour points of the initial contour to obtain a first point set. The convex hull detection module performs convex hull detection on the contour after polygon approximation so that the contour points located on the abdomen of the shrimp are removed from the first point set to form a second point set. The angle contour analysis module performs angle contour analysis on the contour after polygon approximation to form a third point set. The intersection set calculation module is used to obtain the intersection of the second point set and the third point set, and the intersection is the feature set of the shrimp head and shrimp tail.

2. The automatic pond inspection method for crayfish infrared image recognition according to claim 1 is characterized by: In step 1), after the image enhancement module grayscales the image, the pixel value of each pixel in the image is recorded. Then, the image is divided into several partition windows. Each partition window consists of a square composed of an odd number of pixels. The pixels in each square are arranged and the median is taken. Finally, several partition windows with the median taken are combined to obtain the denoised image.

3. The automatic pond inspection method for crayfish infrared image recognition according to claim 1 is characterized in that: In step 2), the binarization module sets the grayscale value of the pixel points on the image after the noise reduction module to 0 or 255 for image binarization, sets the threshold to 158, sets the value of all pixels greater than 158 in the image to 255, and the rest to 0, to obtain a black and white image.

4. The automatic pond inspection method for crayfish infrared image recognition according to claim 1 is characterized in that: In step 3), the edge detection module uses Canny edge detection, performs filter convolution calculation on the image, calculates the gradient amplitude and direction of the image to obtain the total of all possible edges of the image, and then performs non-maximum suppression to eliminate non-edge pixels, retaining only candidate edges of thin lines. Then, double threshold screening selects edges that fall within the threshold range. The double thresholds are set to 85 and 140 respectively, completing the edge processing of the image to form a feature box diagram.

5. A system for patrolling ponds according to the automatic pond patrol method using infrared image recognition of crayfish according to any one of claims 1 to 4, characterized in that: It includes an infrared image acquisition device and an image processing unit arranged on a UAV carrier platform, and the infrared image acquisition device and the image processing unit are wirelessly connected to the interactive platform end; the image processing unit includes an image enhancement module, a binarization module, an edge detection module, a target extraction module and a target counting module connected in sequence; The image enhancement module is used for image noise reduction to filter out noise information in the image; The binarization module performs black and white processing on the image obtained by the noise reduction module to obtain an image with clear feature contours; The edge detection module is used to obtain edge-forming feature frames in an image; The target extraction module is used to detect the number of feature contours; The target counting module outputs the number obtained by the target extraction module.

Images