Automatic sorting and packaging device for navel oranges
By designing an automatic navel orange sorting and packaging device containing multiple sets of image acquisition and auxiliary rolling components, and combining deep learning models for defect identification, the problems of incomplete image acquisition and low sorting in the prior art are solved, and more efficient and accurate navel orange sorting is achieved.
Patent Information
- Application Number
- CN202411984981.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-31
- Publication Date
- 2025-03-25
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing automatic navel orange sorting and packaging device results in low sorting accuracy when the picture collection is not comprehensive and it is inconvenient to turn the navel orange for multiple pictures.
An automatic sorting and packaging device for navel oranges including installation components, conveying components, image acquisition components, auxiliary rolling components and cutting components is designed. By setting up two sets of left and right image acquisition components and auxiliary rolling components, the comprehensive picture acquisition and flip of navel oranges is realized, and defect identification and classification are combined with deep learning models.
It improves the sorting efficiency and defect identification of navel oranges, achieves higher sorting accuracy, and reduces equipment interference and blockage during image collection.
Smart Images
Figure CN119680908A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of automatic sorting of navel oranges, and in particular to an automatic sorting and packaging device for navel oranges. Background Art
[0002] As a popular fruit, navel oranges need to be sorted and packaged after picking to meet different market demands and standards. Traditionally, navel orange sorting and packaging mainly rely on manual operations, where workers observe the defects of navel oranges with their naked eyes. With the development of the times, high-precision image acquisition systems and optimized image recognition algorithms are used to achieve automatic sorting of navel oranges.
[0003] However, the existing related devices still have some deficiencies in structural design and functional realization, and it is difficult to meet the actual needs of efficient and accurate sorting and packaging. The first is the deficiency of image collection. Traditional image collection is to take pictures of navel oranges at multiple fixed side points, and then collect pictures. The collection is not comprehensive (the position where the navel orange contacts the conveying equipment is easily blocked, which can also be understood as the bottom of the navel orange), which can easily affect the subsequent sorting effect. It is inconvenient to turn over the navel orange for multiple picture collection, and the sorting accuracy is low.
[0004] In order to solve the above problems, the present application proposes an automatic sorting and packaging device for navel oranges. Summary of the invention
[0005] The object of the present invention is to provide an automatic sorting and packaging device for navel oranges to solve the problems in the prior art proposed in the above background technology that image collection is not comprehensive, it is inconvenient to turn over the navel oranges for multiple image collection, and the sorting accuracy is low.
[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: an automatic sorting and packaging device for navel oranges, comprising a navel orange classification mechanism and a packaging device, the navel orange classification mechanism comprising a mounting component, a conveying component, an image acquisition component, an auxiliary rolling component, a material discharge component and a processing unit, the mounting component is mounted with the conveying component, the mounting component is mounted with two groups of the image acquisition components, the mounting component is mounted with the auxiliary rolling component, the mounting component is mounted with the material discharge component, the two groups of the image acquisition components are arranged on the left and right, the auxiliary rolling component is arranged between the two image acquisition components, the material discharge component is arranged on the right side of the image acquisition component located at the right end, the mounting component is mounted with the processing unit, and the packaging device is arranged at the lower right side of the conveying component.
[0007] Furthermore, the mounting assembly includes a mounting frame, a baffle and a base, the baffle is fixedly connected to both the front and rear sides of the top of the mounting frame, and the base is fixedly connected to the bottom end of the mounting frame.
[0008] Furthermore, each group of the conveying components includes a motor A, a turntable, a connecting belt and a roller. The turntable is provided on the inner side of the front end surface and the inner side of the rear end surface of the mounting frame. The number of turntables in each group is two. The connecting belts are frictionally connected to the outer sides of the two turntables. The sides of the two connecting belts that are close to each other are rotatably connected to the rollers through a rotating shaft. The sides of the two turntables at the right end that are away from each other are rotatably connected to the mounting frame through a rotating shaft. The sides of the two turntables at the left end that are away from each other are fixedly connected to the output end of the motor A, and one side of the motor A is fixedly connected to the mounting frame.
[0009] Furthermore, the image acquisition component includes a bracket A, a bracket B, a mounting plate, a camera, a ring-shaped fill light and a strip-shaped fill light. The tops of the two baffles are fixedly connected to the bracket A, the front and rear sides of the bottom of the mounting frame are fixedly connected to the bracket B, the ends of the bracket A and the bracket B away from the mounting frame are fixedly connected to the mounting plate, the end of the mounting plate close to the mounting frame is fixedly connected to the camera, the outer side of the camera is provided with the ring-shaped fill light, one side of the ring-shaped fill light is fixedly connected to the corresponding mounting plate, and the sides of the two brackets A close to each other are fixedly connected to the strip-shaped fill light.
[0010] Furthermore, the auxiliary rolling assembly includes a base, an electric telescopic rod, a push plate, a spring, a block and a friction plate. The base is fixedly connected to the inner sides of the front end surface and the rear end surface of the mounting frame. The top of the base is fixedly connected to the electric telescopic rod, the top of the electric telescopic rod is fixedly connected to the push plate, the top of the push plate is fixedly connected to the spring, the top of the spring is fixedly connected to the block, the top of the block is fixedly connected to the friction plate, the motion trajectory of the friction plate is linear, and the roller is arranged on the motion trajectory of the friction plate.
[0011] Furthermore, the unloading assembly includes a fixed frame, a motor B, a rotating drum, a paddle, a mounting bar and a photoelectric sensor. The fixed frame is fixedly connected to the top of the baffle, the motor B is fixedly connected to the top center position of the left side of the fixed frame, the output end of the motor B is fixedly connected to the rotating drum, the four paddles are fixedly connected to the outside of the rotating drum, the mounting bar is fixedly connected to the inside of the fixed frame, the photoelectric sensor is fixedly connected to the inside of the mounting bar, the output end of the motor B passes through the fixed frame, the rotating drum and the photoelectric sensor are arranged directly above the center of the interval between the two groups of conveying assemblies, and the baffle is provided with a discharge hole at the position corresponding to the rotating drum.
[0012] Furthermore, the processing unit includes an image preprocessing module, a feature extraction module, a deep learning model, a classification decision module and a control module. The image preprocessing module performs grayscale processing on the received image, converts the color image into a grayscale image, and then performs noise reduction processing on the grayscale image through a median filtering algorithm to remove salt and pepper noise and Gaussian noise in the image to make the image smoother and clearer. The image is enhanced by a histogram equalization method to expand the grayscale dynamic range of the image and improve the contrast of the image.
[0013] The feature extraction module adopts an extraction method based on texture features. The feature extraction module uses a gray level co-occurrence matrix algorithm to calculate the texture feature parameters of the image, such as contrast, correlation, energy and entropy. The texture features of the defective area are usually significantly different from the texture of the normal navel orange surface. Through the analysis and threshold setting of these texture feature parameters, the defect is classified and evaluated according to the size, shape and severity of the defect.
[0014] The deep learning model is based on a convolutional neural network and is used to identify and classify the comprehensive features of navel oranges. The deep learning model includes multiple convolutional layers, pooling layers and fully connected layers. The convolutional layer extracts features by sliding a convolution kernel on an image. The pooling layer downsamples the feature map to reduce the amount of data while retaining the main features. The fully connected layer is used to integrate and classify the extracted features. In the model training stage, a large number of navel orange image samples of different varieties and different growth environments are collected, including samples of various defects, and are manually annotated and classified. The annotated image samples are divided into a training set, a validation set and a test set. The convolutional neural network model is trained using a back propagation algorithm, and the weight and bias parameters of the model are continuously adjusted so that the model can accurately identify and classify the features of the navel oranges. After multiple iterations of training, the accuracy of the model on the validation set and the test set reaches 99.5%;
[0015] The classification decision module judges and makes decisions on the output results of the model based on the preset classification thresholds and rules and the feature vector of the navel orange output by the deep learning model 163;
[0016] The control module receives the judgment and decision of the classification decision module, and controls the motor B to operate to classify and remove defective navel oranges.
[0017] Compared with the prior art, the present invention has the following beneficial effects:
[0018] The present invention arranges a processing unit, an image acquisition component, an auxiliary rolling component, a material discharge component and a processing unit, and a photoelectric sensor detects the incoming fruit. When the defect size is smaller than a threshold value, the motor B rotates forward, and the paddle rotates to discharge the navel orange forward. When the defect size is larger than the threshold value, the motor B rotates reversely, and the paddle rotates to discharge the navel orange backward. If the fruit is good, the material discharge component does not operate, and the navel orange is transported to the right side by the conveying component and falls on the packaging equipment for collection and packaging, thereby improving the sorting efficiency of the navel orange, improving the recognition of defects, and sampling twice, which is convenient for turning over the navel orange, and reducing the interference and obstruction of other equipment to the navel orange during picture collection, thereby improving the sampling of the picture, and thus improving the accuracy of sorting. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 This is a schematic diagram of the overall structure of an automatic navel orange sorting and packaging device of the present invention;
[0020] Figure 2 It is a schematic diagram of the installation structure of the installation components of an automatic navel orange sorting and packaging device of the present invention;
[0021] Figure 3 It is a schematic diagram of the installation structure of a conveying component of an automatic navel orange sorting and packaging device of the present invention;
[0022] Figure 4 This is a schematic diagram of the installation structure of an image acquisition component of an automatic navel orange sorting and packaging device of the present invention;
[0023] Figure 5 It is a schematic diagram of the installation structure of an auxiliary rolling assembly of an automatic navel orange sorting and packaging device of the present invention;
[0024] Figure 6 It is a schematic diagram of the installation structure of a material unloading component of an automatic navel orange sorting and packaging device of the present invention;
[0025] Figure 7 This is a structural diagram of a processing unit of an automatic navel orange sorting and packaging device of the present invention;
[0026] Figure 8 A flow chart of sorting navel oranges by an automatic navel orange sorting and packaging device according to the present invention;
[0027] Fig. 9 A top view of the overall structure of an automatic navel orange sorting and packaging device according to the present invention;
[0028] In the figure:
[0029] 1. Navel orange classification organization;
[0030] 11. Installation assembly; 111. Installation frame; 112. Baffle; 113. Base;
[0031] 12. Conveying assembly; 121. Motor A; 122. Turntable; 123. Connecting belt; 124. Roller;
[0032] 13. Image acquisition component; 131. Bracket A; 132. Bracket B; 133. Mounting plate; 134. Camera; 135. Ring fill light; 136. Strip fill light;
[0033] 14. Auxiliary rolling assembly; 141. Base; 142. Electric telescopic rod; 143. Push plate; 144. Spring; 145. Stopper; 146. Friction plate;
[0034] 15. blanking assembly; 151. fixing frame; 152. motor B; 153. rotating drum; 154. paddle; 155. mounting strip; 156. photoelectric sensor;
[0035] 16. Processing unit; 161. Image preprocessing module; 162. Feature extraction module; 163. Deep learning model; 164. Classification decision module; 165. Control module;
[0036] 2. Packing equipment. DETAILED DESCRIPTION
[0037] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.
[0038] The standard parts used in this application can all be purchased from the market, and special-shaped parts can be customized according to the instructions and drawings. The specific connection methods of each part adopt conventional means such as mature bolts, rivets, welding, and bonding in the existing technology, and the components used for circuit connection are all conventional models in the existing technology.
[0039] At the same time, in order to clearly express the connection relationship and working principle between the components and highlight the key points, the drawings in the specification are organized and drawn in the form of simple diagrams. One simple diagram can correspond to a variety of materials and actual external structural shapes.
[0040] See also Figure 1 — Fig. 9The present invention provides a technical solution: an automatic sorting and packaging device for navel oranges, comprising a navel orange classification mechanism 1 and a packaging device 2, the navel orange classification mechanism 1 comprising a mounting component 11, a conveying component 12, an image acquisition component 13, an auxiliary rolling component 14, a material discharge component 15 and a processing unit 16, the mounting component 11 is mounted with a conveying component 12, the mounting component 11 is mounted with two groups of image acquisition components 13, the mounting component 11 is mounted with an auxiliary rolling component 14, the mounting component 11 is mounted with a material discharge component 15, the two groups of image acquisition components 13 are arranged on the left and right, the auxiliary rolling component 14 is arranged between the two image acquisition components 13, the material discharge component 15 is arranged on the right side of the image acquisition component 13 at the right end, the mounting component 11 is mounted with a processing unit 16, and a packaging device 2 is arranged below the right side of the conveying component 12.
[0041] The mounting assembly 11 includes a mounting frame 111 , a baffle 112 and a foot 113 . The baffle 112 is fixedly connected to both the front and rear sides of the top of the mounting frame 111 , and the foot 113 is fixedly connected to the bottom of the mounting frame 111 .
[0042] Each group of conveying components 12 includes a motor A121, a turntable 122, a connecting belt 123 and a roller 124. Turntables 122 are provided on the inner sides of the front end surface and the rear end surface of the mounting frame 111. There are two turntables 122 in each group. The outer sides of the two turntables 122 are frictionally connected with connecting belts 123. The sides of the two connecting belts 123 that are close to each other are rotatably connected to the roller 124 through a rotating shaft. The sides of the two turntables 122 at the right end that are away from each other are rotatably connected to the mounting frame 111 through a rotating shaft. The sides of the two turntables 122 at the left end that are away from each other are fixedly connected to the output end of the motor A121. One side of the motor A121 is fixedly connected to the mounting frame 111. The auxiliary rolling component 1 The mounting frame 111 includes a base 141, an electric telescopic rod 142, a push plate 143, a spring 144, a stopper 145 and a friction plate 146. The inner sides of the front end surface and the rear end surface of the mounting frame 111 are fixedly connected to the base 141. The top of the base 141 is fixedly connected to the electric telescopic rod 142. The top of the electric telescopic rod 142 is fixedly connected to the push plate 143. The top of the push plate 143 is fixedly connected to the spring 144. The top of the spring 144 is fixedly connected to the stopper 145. The top of the stopper 145 is fixedly connected to the friction plate 146. The motion track of the friction plate 146 is linear. The roller 124 is arranged on the motion track of the friction plate 146. The navel oranges are placed on the conveying assembly 1 from the left side of the navel orange sorting mechanism 1. 2, the navel orange is located between the two groups of rollers 124, and then the motor A121 runs to drive the left turntable 122 to rotate, thereby causing the connecting belt 123 to start rolling, and the connecting belt 123 drives the right turntable 122 to rotate on the mounting frame 111, and then the navel orange between the rollers 124 is transported by the two groups of conveying components 12 running synchronously, and the navel orange is first imaged by the first image acquisition component 13, and the conveying component 12 continues to transport the navel orange, and when the roller 124 passes the friction plate 146, the roller 124 rolls on the friction plate 146, and the rolling of the roller 124 drives the navel orange to roll, and then the roller 124 under the navel orange is separated from the friction plate 146, so that the navel orange The navel orange is no longer rolling, so that it is turned over and adjusted to a certain direction. The navel orange is subjected to a second image acquisition through the image acquisition component 13 on the right side. The electric telescopic rod 142 on the base 141 is extended and retracted to drive the push plate 143 to rise and fall, thereby the stopper 145 is raised and lowered, and then the friction plate 146 is raised and lowered, and the contact force between the friction plate 146 and the roller 124 is adjusted. The spring 144 provides a certain elastic force to the stopper 145. The friction plate 146 can flexibly contact with the roller 124, which is convenient for adapting to navel oranges of different types and sizes. The electric telescopic rod 142 drives the push plate 143 to rise and fall to improve the stability of the contact between the friction plate 146 and the roller 124, and avoid the friction plate 124 and the roller from being too tight or too loose.
[0043] The image acquisition component 13 includes a bracket A131, a bracket B132, a mounting plate 133, a camera 134, a ring-shaped fill light 135 and a strip-shaped fill light 136. The tops of the two baffles 112 are fixedly connected to the bracket A131, the front and rear sides of the bottom of the mounting frame 111 are fixedly connected to the bracket B132, the ends of the brackets A131 and B132 away from the mounting frame 111 are fixedly connected to the mounting plate 133, the end of the mounting plate 133 close to the mounting frame 111 is fixedly connected to the camera 134, the outer side of the camera 134 is provided with a ring-shaped fill light 135, and the ring-shaped fill light 136 is fixedly connected to the mounting plate 133. One side of the two brackets A131 is fixedly connected to the corresponding mounting plate 133, and the sides of the two brackets A131 close to each other are fixedly connected with strip fill lights 136. The upper and lower cameras 134 collect pictures of the upper and lower sides of the navel orange. The bracket A131 is used to fix the top mounting plate 133, and the bracket B132 is used to fix the bottom mounting plate 133. The strip fill lights 136 are strip-shaped, so that the two strip fill lights 136 can fill light on the front and back sides of the navel orange. The camera 134 is annular, so that the two cameras 134 can fill light on the upper and lower parts of the navel orange respectively, thereby improving the picture collection effect.
[0044] The unloading assembly 15 includes a fixed frame 151, a motor B152, a rotating drum 153, a paddle 154, a mounting bar 155 and a photoelectric sensor 156. The top of the baffle 112 is fixedly connected to the fixed frame 151, and the motor B152 is fixedly connected to the top center of the left side of the fixed frame 151. The output end of the motor B152 is fixedly connected to the rotating drum 153, and four paddles 154 are fixedly connected to the outside of the rotating drum 153. The inner side of the fixed frame 151 is fixedly connected to the mounting bar 155, and the inner side of the mounting bar 155 is fixedly connected to the photoelectric sensor 156. The output end of the motor B152 passes through the fixed frame 151. The rotating drum 153 and the photoelectric sensor 156 are arranged directly above the center of the interval between the two groups of conveying assemblies 12. The baffle 112 is provided with a discharge hole at the position corresponding to the rotating drum 153. The control module 165 receives the classified The decision module 164 judges and makes decisions, and controls the motor B152 to classify and remove defective navel oranges. The navel oranges continue to be transported by the conveying component 12. When the navel oranges pass under the mounting bar 155, the photoelectric sensor 156 detects the incoming fruit. When the defect size is smaller than the threshold, the motor B152 on the fixed frame 151 rotates forward to drive the rotating drum 153 to rotate, and the rotating drum 153 drives the paddle 154 to rotate to discharge the navel oranges forward, and the navel oranges pass through the discharge hole of the front conveying component 12. When the defect size is larger than the threshold, the motor B152 reverses and runs, and the paddle 154 rotates to discharge the navel oranges backward, and the navel oranges pass through the discharge hole of the rear conveying component 12. If it is a good fruit, the unloading component 15 does not operate, and the navel oranges are transported to the right through the conveying component 12 and fall on the packaging device 2 for collection and packaging.
[0045] The processing unit 16 includes an image preprocessing module 161, a feature extraction module 162, a deep learning model 163, a classification decision module 164 and a control module 165. The image preprocessing module 161 performs grayscale processing on the received image, converts the color image into a grayscale image, reduces the amount of data while retaining the main feature information of the image, and improves the efficiency of subsequent processing. Then, the grayscale image is subjected to noise reduction processing by a median filtering algorithm to remove salt and pepper noise and Gaussian noise in the image, making the image smoother and clearer. The median filtering effectively suppresses the interference of noise by sorting the neighborhood pixel values of each pixel point in the image and taking the middle value as the new value of the pixel point, while maintaining the edge and detail information of the image. The image is enhanced by a histogram equalization method, the grayscale dynamic range of the image is expanded, and the contrast of the image is improved. By adjusting the grayscale histogram of the image, the details of the originally darker or brighter areas are made clearer and more visible, thereby enhancing the feature expression of the navel orange and providing a better basis for subsequent feature extraction and recognition.
[0046] The feature extraction module 162 uses an extraction method based on texture features. The feature extraction module 162 uses a gray level co-occurrence matrix algorithm to calculate the texture feature parameters of the image, such as contrast, correlation, energy and entropy. The texture features of the defective area are usually significantly different from the texture of the normal navel orange surface. Through the analysis and threshold setting of these texture feature parameters, the defect is classified and evaluated according to the size, shape and severity of the defect.
[0047] The deep learning model 163 is based on a convolutional neural network and is used to identify and classify the comprehensive features of navel oranges. The deep learning model 163 includes multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer extracts features by sliding the convolution kernel on the image, and the pooling layer downsamples the feature map to reduce the amount of data while retaining the main features. The fully connected layer is used to integrate the extracted features and make classification decisions. In the model training stage, a large number of navel orange image samples of different varieties and different growth environments, including samples of various defects, are collected, and manually labeled and classified. The labeled image samples are divided into a training set, a validation set, and a test set. The convolutional neural network model is trained using a back-propagation algorithm, and the weights and bias parameters of the model are continuously adjusted so that it can accurately identify the features of navel oranges and classify them. After multiple iterations of training, the accuracy of the model on the validation set and the test set reaches 99.5%.
[0048] The classification decision module 164 judges and makes decisions on the output results of the model based on the preset classification thresholds and rules and the feature vector of the navel orange output by the deep learning model 163;
[0049] The control module 165 receives the judgment and decision of the classification decision module 164, and controls the motor B152 to classify and remove the defective navel oranges.
[0050] Working principle:
[0051] When the device is in use, the navel orange is placed on the conveying component 12 from the left side of the navel orange sorting mechanism 1, and the navel orange is located between the two groups of rollers 124. Then the motor A121 is operated to drive the turntable 122 on the left side to rotate, thereby causing the connecting belt 123 to start rolling. The connecting belt 123 drives the turntable 122 on the right side to rotate on the mounting frame 111. Then the navel orange between the rollers 124 is transported by the two groups of conveying components 12 running synchronously. The navel orange is first imaged by the first image acquisition component 13. The upper and lower cameras 134 capture images of the upper and lower sides of the navel orange. The bracket A131 is used to monitor the top mounting plate 133. The bracket B132 is used to fix the bottom mounting plate 133. The strip fill light 136 is strip-shaped, so that the two strip fill lights 136 can fill light on the front and back sides of the navel orange. The camera 134 is ring-shaped, so that the two cameras 134 can fill light on the upper and lower parts of the navel orange respectively, thereby improving the picture collection effect. The conveying component 12 continues to convey the navel orange. When the roller 124 passes through the friction plate 146, the roller 124 rolls on the friction plate 146, and the rolling of the roller 124 drives the navel orange to roll. Then the roller 124 under the navel orange is separated from the friction plate 146, so that the navel orange stops rolling, thereby causing the navel orange to flip and adjust its direction.
[0052] The navel orange is captured for the second time through the image acquisition component 13 on the right side. After the image is captured, it is sent to the processing unit 16. The image preprocessing module 161 grays the image and performs noise reduction processing. The feature extraction module 162 analyzes the defect texture feature parameters and sets the threshold value. Classification and evaluation are performed according to the size, shape and severity of the defect. The deep learning model 163 outputs the feature vector of the navel orange. The classification decision module 164 judges and makes decisions on the output results of the model. The control module 165 receives the judgment and decision of the classification decision module 164 and controls the motor B152 to classify and remove the defective navel oranges. The navel oranges continue to pass through the conveying component. 12 conveyor, when the navel orange passes under the mounting bar 155, the photoelectric sensor 156 detects the incoming fruit. When the defect size is smaller than the threshold value, the motor B152 on the fixed frame 151 rotates forward to drive the rotating drum 153 to rotate, and the rotating drum 153 drives the paddle 154 to rotate to discharge the navel orange forward, and the navel orange passes through the discharge hole of the front conveying component 12. When the defect size is larger than the threshold value, the motor B152 reverses and runs, and the paddle 154 rotates to discharge the navel orange backward, and the navel orange passes through the discharge hole of the rear conveying component 12. If it is a good fruit, the unloading component 15 does not operate, and the navel orange is conveyed to the right side by the conveying component 12 and falls on the packaging device 2 for collection and packaging.
Claims
1. An automatic sorting and packaging device for navel oranges, characterized by: The invention comprises a navel orange sorting mechanism (1) and a packaging device (2). The navel orange sorting mechanism (1) comprises a mounting assembly (11), a conveying assembly (12), an image acquisition assembly (13), an auxiliary rolling assembly (14), a material discharge assembly (15) and a processing unit (16). The conveying assembly (12) is mounted on the mounting assembly (11). Two groups of the image acquisition assemblies (13) are mounted on the mounting assembly (11). The auxiliary rolling assembly (14) is mounted on the mounting assembly (11). The material discharge assembly (15) is mounted on the mounting assembly (11). The two groups of the image acquisition assemblies (13) are arranged on the left and right. The auxiliary rolling assembly (14) is arranged between the two image acquisition assemblies (13). The material discharge assembly (15) is arranged on the right side of the image acquisition assembly (13) at the right end. The processing unit (16) is mounted on the mounting assembly (11). The packaging device (2) is arranged below the right side of the conveying assembly (12).
2. The automatic sorting and packaging device for navel oranges according to claim 1, characterized in that: The mounting assembly (11) comprises a mounting frame (111), a baffle (112) and a base (113); the baffle (112) is fixedly connected to both the front and rear sides of the top end of the mounting frame (111); and the base (113) is fixedly connected to the bottom end of the mounting frame (111).
3. The automatic sorting and packaging device for navel oranges according to claim 1, characterized in that: Each group of the conveying components (12) comprises a motor A (121), a turntable (122), a connecting belt (123) and a roller (124); the turntable (122) is provided on the inner side of the front end surface and the inner side of the rear end surface of the mounting frame (111); the number of the turntables (122) in each group is two; the connecting belts (123) are frictionally connected to the outer sides of the two turntables (122); the sides of the two connecting belts (123) close to each other are rotationally connected to the roller (124) via a rotating shaft; the sides of the two turntables (122) at the right end that are away from each other are rotationally connected to the mounting frame (111) via a rotating shaft; the sides of the two turntables (122) at the left end that are away from each other are fixedly connected to the output end of the motor A (121); and one side of the motor A (121) is fixedly connected to the mounting frame (111).
4. The automatic sorting and packaging device for navel oranges according to claim 1, characterized in that: The image acquisition component (13) comprises a bracket A (131), a bracket B (132), a mounting plate (133), a camera (134), a ring-shaped fill light (135) and a strip-shaped fill light (136); the top ends of the two baffles (112) are fixedly connected to the bracket A (131); the front and rear sides of the bottom end of the mounting frame (111) are fixedly connected to the bracket B (132); the bracket A (131) and the bracket B (132) are separated from the mounting frame (111); One end of each of the brackets A (111) is fixedly connected to the mounting plate (133), one end of the mounting plate (133) close to the mounting frame (111) is fixedly connected to the camera (134), the outer side of the camera (134) is sleeved with the ring-shaped fill light (135), one side of the ring-shaped fill light (135) is fixedly connected to the corresponding mounting plate (133), and the sides of the two brackets A (131) close to each other are fixedly connected to the strip-shaped fill light (136).
5. The automatic sorting and packaging device for navel oranges according to claim 1, characterized in that: The auxiliary rolling assembly (14) comprises a base (141), an electric telescopic rod (142), a push plate (143), a spring (144), a stopper (145) and a friction plate (146); the base (141) is fixedly connected to the inner side of the front end surface and the inner side of the rear end surface of the mounting frame (111); the top end of the base (141) is fixedly connected to the electric telescopic rod (142); the top end of the electric telescopic rod (142) is fixedly connected to the push plate (143); the top end of the push plate (143) is fixedly connected to the spring (144); the top end of the spring (144) is fixedly connected to the stopper (145); the top end of the stopper (145) is fixedly connected to the friction plate (146); the motion track of the friction plate (146) is linear; and the roller (124) is arranged on the motion track of the friction plate (146).
6. The automatic sorting and packaging device for navel oranges according to claim 1, characterized in that: The unloading assembly (15) comprises a fixing frame (151), a motor B (152), a rotating drum (153), a paddle (154), a mounting strip (155) and a photoelectric sensor (156); the fixing frame (151) is fixedly connected to the top of the baffle (112); the motor B (152) is fixedly connected to the center of the top left side of the fixing frame (151); the rotating drum (153) is fixedly connected to the output end of the motor B (152); and four photoelectric sensors (156) are fixedly connected to the outside of the rotating drum (153). The paddle (154) is fixedly connected to the inner side of the fixing frame (151), the mounting strip (155) is fixedly connected to the inner side of the mounting strip (155), the output end of the motor B (152) passes through the fixing frame (151), the rotating drum (153) and the photoelectric sensor (156) are arranged directly above the center of the interval between the two groups of conveying assemblies (12), and the baffle (112) is provided with a discharge hole at a position corresponding to the rotating drum (153).
7. The automatic sorting and packaging device for navel oranges according to claim 1, characterized in that: The processing unit (16) comprises an image preprocessing module (161), a feature extraction module (162), a deep learning model (163), a classification decision module (164) and a control module (165). The image preprocessing module (161) performs grayscale processing on the received image to convert the color image into a grayscale image, and then performs noise reduction processing on the grayscale image by using a median filtering algorithm to remove salt and pepper noise and Gaussian noise in the image to make the image smoother and clearer. The image is enhanced by using a histogram equalization method to expand the grayscale dynamic range of the image and improve the contrast of the image. The feature extraction module (162) adopts an extraction method based on texture features. The feature extraction module (162) uses a gray level co-occurrence matrix algorithm to calculate texture feature parameters of the image, such as contrast, correlation, energy and entropy. The texture features of the defect area are usually significantly different from the texture of the surface of a normal navel orange. By analyzing these texture feature parameters and setting thresholds, classification and evaluation are performed according to the size, shape and severity of the defect. The deep learning model (163) is based on a convolutional neural network and is used to identify and classify the comprehensive features of navel oranges. The deep learning model (163) includes multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer extracts features by sliding a convolution kernel on an image. The pooling layer downsamples the feature map to reduce the amount of data while retaining the main features. The fully connected layer is used to integrate and classify the extracted features. In the model training stage, a large number of navel orange image samples of different varieties and different growth environments are collected, including samples of various defects, and are manually annotated and classified. The annotated image samples are divided into a training set, a validation set, and a test set. The convolutional neural network model is trained using a back-propagation algorithm, and the weight and bias parameters of the model are continuously adjusted so that the model can accurately identify the features of the navel oranges and classify them. After multiple iterations of training, the accuracy of the model on the validation set and the test set reaches 99.5%. The classification decision module (164) judges and makes decisions on the output result of the model based on the preset classification threshold and rules and on the basis that the deep learning model 163 outputs a feature vector of navel orange; The control module (165) receives the judgment and decision of the classification decision module (164), and controls the motor B (152) to operate to classify and remove defective navel oranges.