A smart detection system and method for proportion of pilose antler wax tablets

Abstract

The present application relates to a kind of antler wax piece proportion intelligent detection system and detection method, comprising: industrial camera imaging module, for obtaining the appearance image of antler by industrial camera;X-ray imaging module, for obtaining the X-ray imaging of antler by X-ray;Terminal, for the appearance image of antler is sharp processing, skeleton extraction, by calculating skeleton contour, obtain the final length L of whole antler;X-ray imaging is detected by semantic segmentation model, and the wax piece part of antler is detected, and the size of wax piece is obtained;And according to the final length L of whole antler and wax piece size, the proportion of wax piece in antler is obtained.The present application is based on high-resolution image acquisition and X-ray image acquisition, combined with image processing and deep learning feature extraction, realize the accurate detection and positioning of antler wax piece.

Description

Technical Field

[0001] This invention patent relates to the field of X-ray detection technology, specifically to an intelligent detection system and method for detecting the proportion of deer antler wax slices. Background Technology

[0002] X-ray analysis is a non-destructive testing technique based on the penetrating properties of X-rays and their interaction with matter. This method accurately reflects the internal structural characteristics of an object by measuring the attenuation of X-rays as they penetrate it. In the field of agricultural product quality testing, X-ray analysis has become a research hotspot due to its non-destructive nature and high efficiency. Its core principle lies in the interaction between X-rays and the atoms of a substance through the photoelectric effect and Compton scattering, leading to an attenuation of the radiation intensity. The degree of attenuation is closely related to parameters such as the density and atomic number of the substance. By detecting the intensity changes of transmitted X-rays and combining this with image reconstruction algorithms, a visual representation of the internal structure of an object can be achieved.

[0003] As a traditional and precious Chinese medicinal material, deer antler exhibits significant differences in medicinal and economic value across different parts. Modern processing techniques divide deer antler slices into four typical regions from top to bottom: wax slices, powder slices, gauze slices, and bone slices. Among these, wax slices, located at the top of the antler, are rich in various active ingredients (such as amino acids, polysaccharides, and growth factors), exhibiting outstanding efficacy in promoting tissue repair and enhancing immunity, and commanding a market price dozens of times higher than other parts. Studies have shown that wax slices differ significantly from other parts in physical properties: their tissue structure is denser, their protein content is higher, and the density difference can reach 15%-20%, providing a theoretical basis for X-ray detection.

[0004] Currently, the detection of deer antler wax content relies on manual methods, which suffer from high subjectivity and large errors. The results depend on the experience and subjective judgment of the personnel, and different individuals may have different understandings and classification standards for wax characteristics, easily leading to inconsistent test results for the same batch of deer antlers. This makes it difficult to guarantee accuracy and may even damage the deer antlers. Therefore, there is an urgent need for a rapid, non-destructive, automated detection system and method for deer antler wax content. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent detector for the proportion of wax flakes in deer antlers. Based on X-ray and high-resolution image acquisition, combined with image processing and deep learning feature extraction, it can accurately detect and locate the wax flakes inside deer antlers, comprehensively covering various sizes of deer antlers and improving the comprehensiveness and accuracy of detection. First, an industrial camera is used to acquire images of the entire deer antler, and the total length of the entire antler is calculated based on image processing algorithms. Second, X-ray scan images of the entire deer antler are obtained, and the wax flake portion is segmented using semantic segmentation technology in deep learning. The length range of the wax flakes is then calculated using image processing methods. Finally, the proportion of wax flake length in the entire deer antler is calculated. This invention can help processing enterprises achieve precise grading. By selecting high-quality deer antlers with a higher wax flake proportion (usually less than 10%), the utilization efficiency of raw materials can be significantly improved. Furthermore, establishing quantitative standards based on X-ray features helps to form a scientific deer antler quality evaluation system and regulate market order.

[0006] The technical solution adopted by this invention to achieve the above objectives is: an intelligent detection system for the proportion of deer antler wax flakes, comprising:

[0007] An industrial camera imaging module is used to acquire images of the appearance of deer antlers using an industrial camera.

[0008] X-ray imaging module, used to obtain X-ray images of deer antlers;

[0009] The terminal is used to sharpen the appearance image of the deer antler and extract the skeleton. By calculating the skeleton contour, the final length L of the entire deer antler is obtained. The X-ray imaging is used to detect the waxy part of the deer antler through a semantic segmentation model to obtain the waxy part size. The proportion of waxy parts in the deer antler is obtained based on the final length L of the entire deer antler and the waxy part size.

[0010] The industrial camera imaging module includes a multi-angle light source, a photoelectric sensor, an industrial camera, and a first conveyor belt;

[0011] The multi-angle light source is positioned above the conveyor belt; it includes a central light source and multiple strip light sources distributed around it, the strip light sources being evenly distributed and each strip light source being equidistant from the central light source; it is used to provide the lighting environment required for image acquisition, ensuring that the acquired image is shadow-free;

[0012] The photoelectric sensor is located on both sides of the first conveyor belt and is used to trigger the industrial camera when the deer antler is conveyed by the conveyor belt into the detection range of the photoelectric sensor.

[0013] The industrial camera is positioned above the conveyor belt and is used to take a picture of the antler when the photoelectric sensor is triggered, so as to obtain an image of the antler's appearance.

[0014] The first conveyor belt, made of white matte PU material, is used to transport deer antlers from the industrial camera imaging module to the X-ray imaging module.

[0015] The X-ray imaging module includes an X-ray source, a detector, and a second conveyor belt;

[0016] The X-ray source is used to emit line-scan X-rays when the second conveyor belt is started, so as to perform X-ray scanning of the deer antlers during the conveying process.

[0017] The detector is located on the opposite side of the X-ray source and is used to receive X-ray imaging data obtained after the X-rays pass through the deer antler.

[0018] The second conveyor belt is used to receive the deer antlers transmitted from the industrial camera imaging module after the deer antlers have been photographed by the industrial camera imaging module.

[0019] The second conveyor belt is equipped with a lead-lined shielded detection space, and a lead door is provided at the entrance of the second conveyor belt of the lead-lined shielded detection space. When the deer antler is imaged by the industrial camera imaging module, the lead door is opened. When the deer antler is conveyed into the lead-lined shielded detection space, the lead door is closed, and the X-ray source and detector are activated to acquire X-ray images of the deer antler.

[0020] A method for intelligent detection of the proportion of deer antler wax flakes includes the following steps:

[0021] The industrial camera imaging module acquires images of the appearance of deer antlers using an industrial camera;

[0022] The X-ray imaging module obtains X-ray images of deer antlers using X-rays;

[0023] The terminal sharpens the image of the antler and extracts its skeleton. By calculating the skeleton contour, the final length L of the entire antler is obtained. The X-ray image is processed using a semantic segmentation model to detect the waxy parts of the antler and obtain the waxy part size. The proportion of waxy parts in the antler is obtained based on the final length L of the entire antler and the waxy part size.

[0024] The X-ray imaging module acquires X-ray images of deer antlers using X-rays, including the following steps:

[0025] After the antler is conveyed by the first conveyor belt to the end of the industrial camera imaging module belt, the second conveyor belt in the X-ray imaging module is activated and the X-ray source emits line-scanning X-rays, so that the antler is scanned by X-rays during the conveying process, and the detector receives the X-ray imaging data obtained after the X-rays pass through the antler.

[0026] The process of sharpening and extracting the skeleton from the image of the antler, and then calculating the skeleton contour to obtain the final length L of the entire antler, includes the following steps:

[0027] (1) First, the appearance image of deer antlers is denoised by median filtering;

[0028] (2) Binarize the denoised image so that the background has a pixel value of 255 and the antler has a pixel value of 0.

[0029] (3) Sharpen the head and tail of the deer antlers in the binarized image, as follows:

[0030] 3.1) First, starting from the head to the tail of the binarized antler image, traverse each column of pixels. When the first 0 pixel value (i.e., a black pixel) appears, mark it as the starting point SK, record the column where the starting point is located, and mark the column LE where the distance from the starting point is N pixels. On column LE, find two positions where 0 pixels and 255 pixels alternate from the top to the bottom of the antler, which are the two positions where the background and the antler intersect, U1 and U2. Finally, connect the starting point SK and the two intersecting positions U1 and U2 to form a triangle. Set all pixel values ​​of LE that are not inside the triangle but are biased towards the head to 255, which is the same as the background color, to obtain the final sharpening result.

[0031] 3.2) Flip the current image 180° and repeat step 3.1) to sharpen the beginning and end of the antler.

[0032] (4) Extract the skeleton from the sharpened image to obtain the skeleton of the deer antler;

[0033] (5) Calculate the skeleton contour. Calculate the contour of the multi-cluster pixel set in the skeleton using the contour detection function, find the minimum bounding rectangle of all contours, and sum the lengths of all minimum bounding rectangles to obtain the final length of the skeleton, that is, the final length L of the whole deer antler.

[0034] The X-ray imaging process, using a semantic segmentation model, detects the waxy portion of the antler to obtain its size, and includes the following steps:

[0035] The waxy parts of the antler are detected using a semantic segmentation model, and the areas detected as waxy parts are masked with red.

[0036] Based on the red mask, the antler image is segmented to obtain the wax plate region image of the antler;

[0037] For the cut wax slice region image, the height is first obtained by calculating the minimum bounding rectangle and taken as the maximum length L of the wax slice. maxSecondly, the central axis along the width of the wax sheet is found using PCA principal component analysis, and its normal is plotted. The points where non-red pixels first appear at the two ends of the normal are calculated. The distance between these two points is then calculated to obtain the minimum length L of the wax sheet. min .

[0038] The semantic segmentation model constructs a U-shaped segmentation method for identifying wax slices. 2 -net network and training a pre-obtained network, including the following steps:

[0039] a. Based on pre-acquired X-ray images, the dataset is augmented by horizontal flipping, vertical flipping, 45° rotation, and 90° rotation.

[0040] b. Label the wax slices in the dataset and construct a training dataset;

[0041] c. Load U 2 The -net network is used as a pre-trained model and trained; the loss function is set to binary cross-entropy.

[0042] d. When the loss no longer decreases, training ends, and the trained semantic segmentation model is obtained.

[0043] The percentage of wax flakes in the antler is determined based on the final length L of the entire antler and the size of the wax flakes, as detailed below:

[0044]

[0045] Among them, P min With P max Indicates the minimum and maximum percentage of wax flakes; L represents the length of the entire antler. min L represents the minimum length of the wax sheet. max This indicates the maximum length of the wax sheet.

[0046] The present invention has the following beneficial effects and advantages:

[0047] 1. A high-resolution industrial camera is used to photograph deer antlers. By adjusting the light intensity, color and direction of light through multi-angle light sources, the photoelectric sensor is automatically triggered when the deer antlers pass by, so as to obtain clear and consistent images, which improves the accuracy and automation of deer antler photography.

[0048] 2. By using image denoising, binarization, sharpening, skeleton extraction, and skeleton contour calculation, noise interference in the acquired images is removed, image quality is improved, and image deformation is eliminated to ensure the accuracy of deer antler detection.

[0049] 3. Generate a detailed analysis report on the proportion of antler wax slices, including the length of antlers and wax slices, internal images of antlers, and the proportion of antler wax slices, to help operators analyze and make decisions.

[0050] 4. By combining visual inspection, shape detection, and deep learning methods, it comprehensively covers all aspects of deer antler wax slices, improving the comprehensiveness and accuracy of wax slice detection.

[0051] As can be seen, the embodiments of the present invention use industrial cameras and X-rays to acquire data images of the internal and external conditions of deer antlers, and calculate the proportion of wax flakes on the antlers based on a model. This provides a wider range of detection and evaluation capabilities than existing automatic deer antler quality evaluation technologies, meaning the embodiments of the present invention have more comprehensive detection and evaluation capabilities. Moreover, the detection process is conducted non-destructively, without harming the deer antlers. Furthermore, since the present invention does not rely on manual methods for detection and sorting, it avoids the various drawbacks of manual methods. In summary, the embodiments of the present invention can improve the accuracy, efficiency, and standard consistency of deer antler evaluation and sorting. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the structure of an intelligent detector for the proportion of deer antler wax flakes in one embodiment of the present invention.

[0053] The components include: 1. Multi-angle light source; 2. High-resolution industrial camera; 3. Photoelectric sensor; 4. First conveyor belt; 5. X-ray source; 6. Detector; 7. Second conveyor belt; 8. Hardware host; and 9. Lead-made lifting door.

[0054] Figure 2 This is a schematic diagram of a method for using an intelligent detector for the proportion of deer antler wax flakes in one embodiment of the present invention.

[0055] Figure 3 This is a schematic diagram illustrating the calculation of the wax sheet range using an intelligent detector for the proportion of deer antler wax sheets in one embodiment of the present invention.

[0056] Figure 4 This is a schematic diagram of the sharpening of deer antler images using an intelligent detector for the proportion of deer antler wax flakes in one embodiment of the present invention. Detailed Implementation

[0057] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0058] This invention provides an intelligent detector for the proportion of deer antler wax flakes. The device includes an industrial camera imaging module, an X-ray imaging module, and a hardware host; the hardware host includes an industrial camera image processing module, an X-ray scan image processing module, a software integration processing module, and an intelligent reporting and decision support module, wherein:

[0059] The industrial camera imaging module includes a multi-angle light source 1, a photoelectric sensor 2, a high-resolution industrial camera 3, and a first conveyor belt 4.

[0060] The multi-angle light source consists of four strip light sources distributed around the perimeter and a ring light source located in the center. It is used to provide the light source environment required for appearance image acquisition, ensuring the consistency of the light source environment and the absence of shadows during appearance image acquisition.

[0061] The photoelectric sensor is used to be triggered when the antler passes by the photoelectric sensor;

[0062] The high-resolution industrial camera is a 1200W pixel color industrial camera with a 6mm lens. It is used to take pictures of the deer antler when the photoelectric sensor is triggered, so as to obtain the appearance information of the deer antler.

[0063] The conveyor belt can be made of white matte PU material and is used to transport deer antlers from the industrial camera imaging module to the X-ray imaging module.

[0064] The X-ray imaging module includes a line-scan X-ray source 5, a detector 6, and a second conveyor belt 7.

[0065] The line scan X-ray source is used to emit line scan X-rays when the conveyor belt starts, and X-ray scanning can be performed during the movement.

[0066] The detector is used to receive X-ray imaging data obtained after the X-rays pass through the deer antler on the opposite side of the X-ray source;

[0067] The conveyor belt can be used to receive deer antlers conveyed from an industrial camera imaging module;

[0068] Hardware host 8 is used to process industrial camera imaging and X-ray imaging.

[0069] The conveyor belt features a lead-lined shielded inspection space. This inspection area, constructed with a "steel plate + lead plate + steel plate" shielding material, is a long, narrow channel. Both ends of the channel are equipped with sealable protective lifting doors (lead doors). The protective lifting doors are controlled by lead doors 9. After the high-resolution industrial camera finishes capturing images of the deer antler, the lifting door opens; when the lifting door closes, the X-ray detector operates to acquire X-ray images of the deer antler. This ensures no radiation leakage during X-ray acquisition and guarantees the safety of personnel.

[0070] See Figure 1 In one embodiment, the intelligent detector for the proportion of deer antler wax flakes includes the following steps:

[0071] (1) The deer antlers are manually placed onto the belt of the industrial camera imaging module;

[0072] (2) Press the start detection button, and the industrial camera imaging module belt will start to perform visual inspection;

[0073] (3) When the X-ray imaging module belt reaches the end of the industrial camera imaging module belt, the X-ray imaging module belt starts, the lifting door opens, the deer antler enters the X-ray machine, and after passing through the entrance lifting door, the X-ray imaging module belt stops and the lifting door closes.

[0074] (4) Once the lifting doors are fully closed, the X-ray imaging module belt automatically starts, and the X-ray system is activated.

[0075] Perform X-ray inspection;

[0076] (5) When the deer antler reaches the inside of the exit lifting door, the X-ray is turned off, the belt stops, and the lifting door opens automatically.

[0077] like Figure 2 As shown, in one embodiment, the industrial camera image processing module includes image denoising, binarization, sharpening, skeleton extraction, and skeleton contour calculation. This includes the following steps:

[0078] (1) First, noise reduction is performed using median filtering. The median filtering noise reduction formula is: I'(x,y)=median{I(i,j)|(i,j)€W(x,y)}, where:

[0079] W(x,y) represents the coordinates of all pixels within a window centered at pixel (x,y); i and j represent the row and column numbers, respectively.

[0080] The `median{·}` option represents taking the median of the pixel values ​​within the window.

[0081] I'(x,y) is the filtered image pixel value.

[0082] (2) Binarize the denoised image so that the background has a pixel value of 255 and the antler has a pixel value of 0.

[0083] (3) Sharpen the head and tail of the deer antlers in the binarized image. A schematic diagram of the processing is shown below. Figure 4 As shown, the details are as follows:

[0084] 3.1) First, starting from the head to the tail of the binarized antler image, traverse each column of pixels. When the first 0 pixel value (i.e., a black pixel) appears, mark it as the starting point SK. Record the column where the starting point pixel is located, and mark the column LE where the value 200 pixels away from the starting point is located. Find two positions where 0 pixels and 255 pixels alternate from the top to the bottom of the antler, which are the two positions where the background and the antler intersect, U1 and U2. Finally, connect the starting point and the two intersecting positions to form a triangle. Set all pixel values ​​of LE that are not inside the triangle but are biased towards the head to 255, which is the same as the background color. The final sharpening result is obtained.

[0085] 3.2) Flip the current image 180° and repeat step 3.1) to sharpen the beginning and end of the antler.

[0086] (4) The skeleton is extracted from the sharpened image. By performing multiple erosion and opening operations on the image, the skeleton of the deer antler is finally obtained.

[0087] (5) After the skeleton extraction is completed, the skeleton contour is calculated. The contour of the multi-cluster pixel set in the skeleton is calculated by the findContours method in OpenCV, and the minimum bounding rectangle of all the found contours is calculated. The lengths of all the minimum bounding rectangles are summed to obtain the final length of the skeleton, which also represents the final length L of the whole deer antler.

[0088] In one embodiment, the X-ray scanning image processing module refers to constructing a U-shaped image processing module for identifying wax sheets based on X-ray images. 2 -net network; using a trained semantic segmentation model, detect the waxy portion of the antler in the new sample, and mask the detected waxy portions with red; based on the red mask, segment the waxy portions. The process includes the following steps:

[0089] (1) Based on X-ray images, construct a U-shaped structure for identifying wax slides. 2 -net network. Includes the following steps:

[0090] a. Augment the dataset by horizontal flipping, vertical flipping, and rotation at 45° and 90°.

[0091] b. Use labelme to label the wax slices in the dataset, construct a training dataset, and divide it according to an 8:2 ratio.

[0092] c. Load the pre-trained model and begin training the semantic segmentation model. The model loss function is set to binary cross-entropy, calculated as follows: where y represents the true label and represents the predicted label. The optimizer is Adam, and the learning rate is set to a fixed-step decay. The learning rate halves every 50 generations, with an initial learning rate of 0.01.

[0093] d. After training for 200 generations, the loss no longer decreases, and the model training ends.

[0094] (2) The waxy parts of the new sample deer antler are detected by the trained semantic segmentation model, and the parts detected as waxy parts are masked with red.

[0095] (3) The image of the antler is segmented to obtain the waxy region image of the antler, such as... Figure 3 As shown;

[0096] For the segmented image, the height is first obtained by calculating the minimum bounding rectangle and then used as the maximum length L of the wax sheet.max Secondly, the central axis along the width of the wax sheet is found using PCA principal component analysis, and its normal is plotted. The points where non-red pixels first appear at the two ends of the normal are calculated. The distance between these two points is then calculated to obtain the minimum length L of the wax sheet. min .

[0097] In one embodiment, the software integration processing module includes calculation of the proportion of deer antler wax slices and intelligent data storage.

[0098] The calculation of the proportion of deer antler wax slices is performed by integrating data from the industrial camera image processing module and the X-ray scan image processing module using software. This includes the following steps:

[0099] By combining the total length of the antler calculated in the industrial camera image processing module and the length range of the antler wax slices calculated in the X-ray scan image processing module, the percentage range of wax slices in a single antler can be calculated. The calculation formula is as follows: Among them, P min With P max This indicates the minimum and maximum percentage of wax flakes. L represents the length of the entire antler. min L represents the minimum length of the wax sheet. max This indicates the maximum length of the wax sheet.

[0100] The intelligent data storage includes data indexing, query optimization, and data backup functions, which facilitates data storage and efficient data retrieval.

[0101] In one embodiment, the intelligent reporting and decision support module includes a wax analysis report.

[0102] The wax slice analysis report is an automatically generated detailed analysis report of the antler wax slice ratio detection results, including the length of the antler and wax slices, internal images of the antler, and the proportion of the antler wax slices, to help operators analyze and make decisions.

[0103] In summary, compared with existing manual evaluation of the proportion of wax flakes in deer antlers based on visual inspection and experience, the embodiments of this invention offer a more comprehensive detection and evaluation capability and a more scientific theoretical basis for assessing the internal and external quality of deer antlers. Specifically, industrial cameras and X-rays are used to acquire data images of the internal and external condition of the deer antlers, and the proportion of wax flakes is calculated based on a model, resulting in a wider detection and evaluation range than existing deer antler quality evaluation technologies. Moreover, the detection process is conducted non-destructively, without harming the deer antlers. Since this invention does not rely on manual detection and sorting, it avoids the various drawbacks of manual methods. In conclusion, the embodiments of this invention can improve the accuracy, efficiency, and standard consistency of deer antler evaluation.

[0104] The various embodiments in this specification are described in a progressive manner. As the device embodiments are basically similar to the method embodiments, the description is relatively simple. For relevant parts, please refer to the description of the method embodiments.

[0105] Those skilled in the art will recognize that, in the above examples, the functions described in this invention can be implemented using hardware, software, widgets, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium.

[0106] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart detection system for the proportion of deer antler wax flakes, characterized in that, include: An industrial camera imaging module is used to acquire images of the appearance of deer antlers using an industrial camera. X-ray imaging module, used to obtain X-ray images of deer antlers; The terminal is used to sharpen the appearance image of the deer antler and extract the skeleton. By calculating the skeleton contour, the final length L of the entire deer antler is obtained. The X-ray imaging is used to detect the wax part of the deer antler through a semantic segmentation model to obtain the wax size. The proportion of wax in the deer antler is obtained based on the final length L of the entire deer antler and the wax size. The process of sharpening and extracting the skeleton from the antler's appearance image, and calculating the skeleton contour to obtain the final length L of the entire antler, is configured to be executed as follows: (1) First, the appearance image of deer antlers is denoised by median filtering; (2) Binarize the denoised image so that the background has a pixel value of 255 and the antler has a pixel value of 0. (3) Sharpen the head and tail of the deer antlers in the binarized image, as follows: 3.1) First, starting from the head to the tail of the binarized antler image, traverse each column of pixels. When the first 0 pixel value (i.e., a black pixel) appears, mark it as the starting point SK, record the column where the starting point is located, and mark the column LE where the distance from the starting point is N pixels. On column LE, find two positions where 0 pixels and 255 pixels alternate from the top to the bottom of the antler, which are the two positions where the background and the antler intersect, U1 and U2. Finally, connect the starting point SK and the two intersecting positions U1 and U2 to form a triangle. Set all pixel values ​​of LE that are not inside the triangle but are biased towards the head to 255, which is the same as the background color, to obtain the final sharpening result. 3.2) Flip the current image 180° and repeat step 3.1) to sharpen the beginning and end of the antler. (4) Extract the skeleton from the sharpened image to obtain the skeleton of the deer antler; (5) Calculate the skeleton contour. Calculate the contour of the multi-cluster pixel set in the skeleton using the contour detection function, find the minimum bounding rectangle of all contours, and sum the lengths of all minimum bounding rectangles to obtain the final length of the skeleton, that is, the final length L of the whole deer antler.

2. The intelligent detection system for the proportion of deer antler wax flakes according to claim 1, characterized in that, The industrial camera imaging module includes a multi-angle light source, a photoelectric sensor, an industrial camera, and a first conveyor belt; The multi-angle light source is positioned above the conveyor belt; it includes a central light source and multiple strip light sources distributed around it, the strip light sources being evenly distributed and each strip light source being equidistant from the central light source; it is used to provide the lighting environment required for image acquisition, ensuring that the acquired image is shadow-free; The photoelectric sensor is located on both sides of the first conveyor belt and is used to trigger the industrial camera when the deer antler is conveyed by the conveyor belt into the detection range of the photoelectric sensor. The industrial camera is positioned above the conveyor belt and is used to take a picture of the antler when the photoelectric sensor is triggered, so as to obtain an image of the antler's appearance. The first conveyor belt, made of white matte PU material, is used to transport deer antlers from the industrial camera imaging module to the X-ray imaging module.

3. The intelligent detection system for the proportion of deer antler wax flakes according to claim 1, characterized in that, The X-ray imaging module includes an X-ray source, a detector, and a second conveyor belt; The X-ray source is used to emit line-scan X-rays when the second conveyor belt is started, so as to perform X-ray scanning of the deer antlers during the conveying process. The detector is located on the opposite side of the X-ray source and is used to receive X-ray imaging data obtained after the X-rays pass through the deer antler. The second conveyor belt is used to receive the deer antlers transmitted from the industrial camera imaging module after the deer antlers have been photographed by the industrial camera imaging module.

4. The intelligent detection system for the proportion of deer antler wax flakes according to claim 3, characterized in that, The second conveyor belt is equipped with a lead-lined shielded detection space, and a lead door is provided at the entrance of the second conveyor belt of the lead-lined shielded detection space. When the deer antler is imaged by the industrial camera imaging module, the lead door is opened. When the deer antler is conveyed into the lead-lined shielded detection space, the lead door is closed, and the X-ray source and detector are activated to acquire X-ray images of the deer antler.

5. A method for intelligent detection of the proportion of deer antler wax flakes, characterized in that, Includes the following steps: The industrial camera imaging module acquires images of the appearance of deer antlers using an industrial camera; The X-ray imaging module obtains X-ray images of deer antlers using X-rays; The terminal sharpens the appearance image of the deer antler and extracts the skeleton. By calculating the skeleton contour, the final length L of the entire deer antler is obtained. The X-ray image is processed by a semantic segmentation model to detect the waxy part of the deer antler and obtain the waxy part size. The proportion of waxy parts in the deer antler is obtained based on the final length L of the entire deer antler and the waxy part size. The process of sharpening and extracting the skeleton from the image of the antler, and then calculating the skeleton contour to obtain the final length L of the entire antler, includes the following steps: (1) First, the appearance image of deer antlers is denoised by median filtering; (2) Binarize the denoised image so that the background has a pixel value of 255 and the antler has a pixel value of 0. (3) Sharpen the head and tail of the deer antlers in the binarized image, as follows: 3.1) First, starting from the head to the tail of the binarized antler image, traverse each column of pixels. When the first 0 pixel value (i.e., a black pixel) appears, mark it as the starting point SK, record the column where the starting point is located, and mark the column LE where the distance from the starting point is N pixels. On column LE, find two positions where 0 pixels and 255 pixels alternate from the top to the bottom of the antler, which are the two positions where the background and the antler intersect, U1 and U2. Finally, connect the starting point SK and the two intersecting positions U1 and U2 to form a triangle. Set all pixel values ​​of LE that are not inside the triangle but are biased towards the head to 255, which is the same as the background color, to obtain the final sharpening result. 3.2) Flip the current image 180° and repeat step 3.1) to sharpen the beginning and end of the antler. (4) Extract the skeleton from the sharpened image to obtain the skeleton of the deer antler; (5) Calculate the skeleton contour. Calculate the contour of the multi-cluster pixel set in the skeleton using the contour detection function, find the minimum bounding rectangle of all contours, and sum the lengths of all minimum bounding rectangles to obtain the final length of the skeleton, that is, the final length L of the whole deer antler.

6. The intelligent detection method for the proportion of deer antler wax flakes according to claim 5, characterized in that, The X-ray imaging module acquires X-ray images of deer antlers using X-rays, including the following steps: After the antler is conveyed by the first conveyor belt to the end of the industrial camera imaging module belt, the second conveyor belt in the X-ray imaging module is activated and the X-ray source emits line-scanning X-rays, so that the antler is scanned by X-rays during the conveying process, and the detector receives the X-ray imaging data obtained after the X-rays pass through the antler.

7. The intelligent detection method for the proportion of deer antler wax flakes according to claim 5, characterized in that, The X-ray imaging process, using a semantic segmentation model, detects the waxy portion of the antler to obtain its size, and includes the following steps: The waxy parts of the antler are detected using a semantic segmentation model, and the areas detected as waxy parts are masked with red. Based on the red mask, the antler image is segmented to obtain the wax plate region image of the antler; For the cut wax slice region image, the height is first obtained by calculating the minimum bounding rectangle and then used as the maximum length of the wax slice. Secondly, the central axis along the width of the wax sheet is found using PCA principal component analysis, and its normal is plotted. The points where non-red pixels first appear at the two ends of the normal are calculated. The distance between these two points is then calculated to obtain the minimum length of the wax sheet. .

8. The intelligent detection method for the proportion of deer antler wax flakes according to claim 7, characterized in that, The semantic segmentation model constructs a U-shaped segmentation method for identifying wax slices. 2 -net network and training a pre-obtained network, including the following steps: a. Based on pre-acquired X-ray images, the dataset is augmented by horizontal flipping, vertical flipping, 45° rotation, and 90° rotation. b. Label the wax slices in the dataset and construct a training dataset; c. Load U 2 The -net network is used as a pre-trained model and trained; the loss function is set to binary cross-entropy. d. When the loss no longer decreases, training ends, and the trained semantic segmentation model is obtained.

9. The intelligent detection method for the proportion of deer antler wax flakes according to claim 5, characterized in that, The percentage of wax flakes in the antler is determined based on the final length L of the entire antler and the size of the wax flakes, as detailed below: ， ； in, and Indicates the minimum and maximum percentage of wax flakes; This indicates the length of the entire deer antler. Indicates the minimum length of the wax sheet. This indicates the maximum length of the wax sheet.

Images