Dynamic grading different-color grain rice reverse selection control system and dynamic grading different-color grain rice reverse selection control method

Abstract

The invention relates to the technical field of rice color selection control, in particular to a dynamically-graded different-color rice reverse selection control system and method.The method comprises the steps that a vibration motor and a guide rail are used for enabling single rice grains to fall, and an industrial camera is used for collecting images; performing local contrast adjustment and dynamic range adjustment on the image based on a chromatic aberration enhancement model of the generative adversarial network; establishing a rice standard chromaticity library, discriminating different-color grain rice by using a dynamic reference chromatography two-way recognition algorithm and a relative color difference mapping strategy, and constructing a different-color grain feature reverse index mechanism to carry out two-way recognition; according to the recognition result, controlling an airflow nozzle to blow the rice grains with different colors into a container for the rice grains with different colors; and after the current batch is separated, the rice fed back into the different-color grain container is recognized and separated again, and normal rice is separated from the different-color grains. According to the method, adaptive bidirectional discrimination and accurate separation of rice color sorting are realized.

Description

Technical Field

[0001] The present application relates to the technical field of rice color sorting control, and specifically discloses a dynamic classification counter-selection control system and method for heterochromatic rice grains. Background Art

[0002] During rice processing, it is necessary to separate and remove off-color rice grains to improve the quality and grade of the rice. Currently, traditional rice color sorting methods mainly include photoelectric colorimetric sorting and image recognition sorting, but these traditional methods have many problems and shortcomings in practical applications.

[0003] Traditional rice color sorting equipment typically uses a one-way discrimination method, meaning it can only identify off-color rice kernels but cannot effectively distinguish normal rice mixed in with off-color kernels. This results in a large amount of mis-separated normal rice being unable to be reverse-sorted, resulting in unnecessary waste. Furthermore, existing color sorting technology lacks the ability to adaptively adjust to fluctuations in rice quality between batches, making it difficult to dynamically adapt to changes in raw material quality.

[0004] Furthermore, in the actual rice color sorting process, precise identification and separation control are difficult to achieve due to the rice's rapid falling speed and high density. The commonly used photoelectric colorimetry method is easily affected by lighting conditions and the rice's posture, and cannot identify and separate rice with the same light transmittance but different colors. Image recognition methods also require high-quality captured images, and recognition speed and accuracy need to be improved.

[0005] Therefore, there is an urgent need for a rice color sorting method that can achieve adaptive separation control and counter-selection of different-colored grains to improve the rice yield and quality, reduce loss and waste, and adapt to quality fluctuations of different batches of raw materials.

[0006] In view of this, the present application proposes a dynamic classification counter-selection control system and method for heterochromatic rice grains. Summary of the Invention

[0007] To achieve the above objectives, the present application provides a dynamic classification system and method for the reverse selection of rice with different colored grains. The specific technical solutions are as follows:

[0008] A dynamic classification method for reverse selection and control of rice with different-colored grains, comprising:

[0009] The rice on the conveyor belt is vibrated by a vibration motor, and a guide rail is set at the end of the conveyor belt to make the single rice grains fall off the conveyor belt;

[0010] The rice falling trajectory is fitted into a trajectory surface and divided into a collection area, a recognition area and a separation area according to the trajectory surface. An industrial camera is used to continuously capture images of falling rice in the collection area.

[0011] The color difference enhancement model based on the generative adversarial network is used to adjust the local contrast of the collected falling rice image, and the dynamic range of the collected falling rice image is adjusted in the RGB and HSI dual domains;

[0012] A rice standard colorimetric library was established, and a dynamic reference chromatogram bidirectional recognition algorithm and relative color difference mapping strategy were used to identify off-color rice grains. Calibration was performed within the recognition area, and a reverse indexing mechanism for off-color grain features was constructed to perform bidirectional identification between off-color rice grains and normal rice.

[0013] The movement trajectory of the heterochromatic rice grains in the separation zone is calculated based on the calibration results, and the opening timing of the air flow nozzle is controlled to blow the heterochromatic rice grains into the heterochromatic rice grain container;

[0014] After separating the current batch of rice with different colored grains, the rice in the container of different colored grains is returned to the entrance of the conveyor belt, and the vibration parameters applied by the vibration motor to the conveyor belt and the conveying speed of the conveyor belt are readjusted. The normal rice is then identified and separated by airflow using the reverse indexing mechanism of the different colored grain characteristics to separate the normal rice from the rice with different colored grains.

[0015] Preferably, the rice to be sorted is conveyed onto a vibrating conveyor belt, and the conveyor belt is vibrated to increase the distance between rice grains on the conveyor belt;

[0016] A servo vibration motor is installed below the conveyor belt, and the servo vibration motor applies controllable vibration to the conveyor belt, and the controllable vibration includes adjustable vibration frequency and vibration amplitude. When the rice moves to the end of the conveyor belt, the guide rail device realizes the separation of the rice into single grains.

[0017] The surface roughness, effective working width and length of the conveyor belt are configured according to production requirements.

[0018] Preferably, the motion trajectories of all rice grains falling from the conveyor belt are fitted in three-dimensional space to form a trajectory surface, and based on the fitted trajectory surface, the trajectory surface is divided into a plurality of functional areas along the falling direction;

[0019] The functional areas include a collection area, an identification area and a separation area; the collection area is arranged in the initial falling section where the rice grains leave the conveyor belt guide rail.

[0020] Preferably, a high-speed linear array CCD camera is used in the collection area to collect images of the falling rice during the falling process. The high-speed linear array CCD camera is installed in the normal direction of the section of the trajectory surface of the collection area, and the optical axis of the camera maintains a fixed distance from the rice grain motion trajectory surface.

[0021] The high-speed linear array CCD camera uses a high-frequency flash LED array for lighting during shooting, which is synchronized with the camera exposure. The LED array is distributed in a ring around the lens of the high-speed linear array CCD camera. The lighting intensity of the LED array is dynamically adjusted through pulse width modulation, and the flash frequency is synchronized with the camera line frequency.

[0022] Preferably, a color difference enhancement model is constructed, wherein the color difference enhancement model is constructed using an improved conditional generative adversarial network architecture, wherein the generative adversarial network architecture includes a generator network and a discriminator network;

[0023] The generator network includes an encoder and a decoder. The encoder adopts a progressive convolutional layer structure, and the decoder reconstructs the enhanced falling rice image through deconvolution operations and introduces jump connections in each layer. The discriminator network adopts a PatchGAN structure to divide the input falling rice image into multiple local areas for authenticity discrimination.

[0024] Preferably, local contrast amplification is performed on the falling rice image processed by the chromatic aberration enhancement model, wherein the local contrast amplification adopts an adaptive histogram equalization technology, and a dynamic block strategy is designed according to the color distribution characteristics of the falling rice image, so that each falling rice image is divided into sub-blocks, and the area size of the sub-block is adaptively determined according to the size of the rice grain;

[0025] The falling rice image is subjected to local contrast amplification and then processed simultaneously in RGB and HSI color spaces using a dual-domain color shift stretching technique. In the RGB domain, dynamic range stretching is performed independently on each color channel, and in the HSI domain, hue and saturation components are enhanced.

[0026] Preferably, build rice standard chromaticity library as rice identification benchmark, described standard chromaticity library obtains chromaticity parameter by gathering normal rice sample, extracts chromaticity feature vector in CIELab color space to each sample, and chromaticity parameter is divided into multiple standard chromaticity intervals by cluster analysis;

[0027] A dynamic reference chromatogram bidirectional recognition algorithm is constructed. The dynamic reference chromatogram bidirectional recognition algorithm extracts chromaticity feature vectors from the enhanced falling rice image, calculates the color difference distance between the chromaticity feature vector extracted from the falling rice image and each reference point in the standard chromaticity library, and constructs a dynamic threshold function based on the color difference distribution. The dynamic threshold function automatically adjusts the discrimination standard according to the overall chromaticity characteristics of different batches of rice;

[0028] A relative color difference mapping strategy was constructed to identify normal rice and rice with different colored grains by constructing a two-dimensional color difference distribution map.

[0029] Preferably, the abnormal grain rice in the identification area is calibrated in real time, and the spatial coordinates of each abnormal grain rice in the identification area are obtained through laser positioning after the abnormal grain rice is calibrated, and the trajectory of the abnormal grain rice in the identification area is calculated, and the trajectory of the abnormal grain rice in the separation area is calculated according to the trajectory of the abnormal grain rice in the identification area.

[0030] Preferably, an abnormal grain feature reverse index mechanism is established, the chroma feature of each rice judged as abnormal grain rice is recorded, and a comprehensive feature vector of the abnormal grain rice including texture, shape and local color distribution is constructed;

[0031] The normal rice in the collected abnormal grain rice is counter-selected through the abnormal grain feature reverse index mechanism, and the normal rice in the abnormal grain rice is separated.

[0032] Preferably, the abnormal grain rice in the falling rice is blown out in the separation area by airflow, and the airflow is blown out from the airflow nozzle corresponding to the separation area;

[0033] The airflow nozzles are arranged in an array, a plurality of airflow nozzles are uniformly distributed in the separation area in the transverse direction, each airflow nozzle is controlled to open and close by a high-speed electromagnetic valve, and the effective range of the airflow nozzle is determined by fluid dynamics simulation;

[0034] The control strategy of the airflow nozzle adopts a three-level regulation mechanism of prediction-trigger-feedback, the actual arrival position of the rice is detected by a laser sensor array in the trigger stage, and when the deviation between the detection signal and the predicted time is less than a set threshold, the corresponding nozzle is started to blow the abnormal grain rice into the abnormal grain rice container.

[0035] Preferably, the rice collected in the abnormal grain rice container is re-injected into the inlet of the vibrating conveyor belt, and the vibration parameters and running speed of the conveyor belt are optimized and adjusted to expand the spacing of the rice on the conveyor belt;

[0036] When the rice passes through the identification area again, the abnormal grain feature reverse index mechanism is started to identify the normal rice, the comprehensive feature vector of each rice is extracted, and matching identification is performed in the reverse index table;

[0037] In the counter-selection process, the normal rice is blown out by the airflow nozzle in the separation area of the falling curved surface, so that the normal rice is deflected into the collection container during the falling process.

[0038] A dynamic grading abnormal grain rice counter-selection control system for the dynamic grading abnormal grain rice counter-selection control method, comprising: a vibrating conveying module, a falling separation module, an image enhancement module, a rice grain identification module, a pneumatic separation module and a counter-selection module.

[0039] The vibration conveyor module applies vibration to the rice on the conveyor belt through a vibration motor, and a guide rail is provided at the end of the conveyor belt to allow single rice grains to fall off the conveyor belt;

[0040] The falling rice partitioning module fits the rice falling trajectory into a trajectory surface and divides the falling rice into a collection area, an identification area and a separation area according to the trajectory surface, and continuously collects falling rice images in the collection area using an industrial camera;

[0041] The image enhancement module performs local contrast adjustment on the collected falling rice image based on the color difference enhancement model of the generative adversarial network, and performs dynamic range adjustment on the collected falling rice image in the RGB and HSI dual domains;

[0042] The rice grain recognition module establishes a rice standard colorimetric library, utilizes a dynamic reference color spectrum bidirectional recognition algorithm and a relative color difference mapping strategy to discriminate heterochromatic rice, and performs calibration within the recognition area. It also constructs a heterochromatic grain feature reverse indexing mechanism to perform bidirectional identification of heterochromatic rice and normal rice.

[0043] The pneumatic separation module calculates the movement trajectory of the heterochromatic rice grains in the separation zone according to the calibration results, and controls the opening timing of the air flow nozzle to blow the heterochromatic rice grains into the heterochromatic rice container;

[0044] After separating the current batch of rice with different-colored grains, the reverse selection module returns the rice in the container with different-colored grains to the entrance of the conveyor belt, readjusts the vibration parameters applied by the vibration motor to the conveyor belt and adjusts the conveying speed of the conveyor belt, and then uses the reverse indexing mechanism of the different-colored grain characteristics to identify the normal rice and separate it with airflow, thereby separating the normal rice from the rice with different-colored grains.

[0045] Beneficial effects of the present application: The present application realizes the single-grain falling of rice by setting a vibration motor and a guide rail, providing good conditions for subsequent image acquisition and recognition. The mechanical preprocessing method is simple, reliable and easy to control, and has good stability and reliability.

[0046] This application accurately divides the collection area, recognition area and separation area by fitting the falling trajectory of rice, and then uses an industrial camera to continuously collect images, providing a data basis for subsequent image processing and discrimination, and ensuring the real-time and accuracy of the collected data.

[0047] This application adopts a color difference enhancement model based on a generative adversarial network and dynamic range adjustment of RGB and HSI dual domains, which effectively improves the quality of the collected images, enhances color contrast and local details; it is beneficial to improve the accuracy of subsequent identification of heterochromatic rice grains, especially for heterochromatic grains with small color differences, which can significantly improve the recognition effect.

[0048] This application constructs a rice standard colorimetry library and adopts a dynamic reference chromatogram bidirectional recognition algorithm and a relative color difference mapping strategy to accurately identify heterochromatic rice grains. The bidirectional recognition mechanism can not only identify heterochromatic rice grains, but also identify normal rice among heterochromatic grains through reverse indexing.

[0049] According to the recognition results, the present application accurately controls the opening timing of the airflow nozzle, which can efficiently and accurately separate the rice grains with different colors; the airflow separation response speed is fast, which is conducive to improving the rice sorting efficiency.

[0050] The present application can minimize the loss in the rice screening process by returning the off-color grains and re-adjusting the vibration and transmission parameters, and then using the reverse indexing mechanism for secondary screening, thereby ensuring the removal of the off-color grains while minimizing the loss as much as possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flow chart of a dynamic classification and counter-selection control method for heterochromatic rice grains provided in this application;

[0052] Figure 2 Provide a flow chart of rice delivery and single grain shedding for this application;

[0053] Figure 3 Provide a flow chart of rice falling area division and image acquisition for this application;

[0054] Figure 4 Provide a flow chart of rice image chromatic aberration enhancement and dynamic range stretching for this application;

[0055] Figure 5 A flow chart for identifying and calibrating rice with different colored grains is provided for this application;

[0056] Figure 6 A schematic diagram of the calibration results of rice with different colored grains is provided for this application;

[0057] Figure 7 Provide a flow chart for the separation of different-colored rice grains;

[0058] Figure 8 Provide a normal rice reverse selection flow chart for this application;

[0059] Figure 9 This is a structural diagram of a dynamic grading and counter-selection control device for heterochromatic rice grains provided in this application. DETAILED DESCRIPTION

[0060] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the drawings in the specification.

[0061] In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can also be implemented in other ways different from this description. Those skilled in the art can make similar generalizations without violating the connotation of the present application. Therefore, the present application is not limited to the specific embodiments disclosed below.

[0062] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present application. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it constitute a separate or selective embodiment that is mutually exclusive with other embodiments.

[0063] Example 1

[0064] Reference Figures 1 to 8 , which is the first embodiment of this application, such as Figure 1 As shown, a flow chart of a dynamic classification and control method for reverse selection of heterochromatic rice grains is provided.

[0065] Step 1: Use a vibration motor to vibrate the rice on the conveyor belt, and set a guide rail at the end of the conveyor belt to make the rice grains fall off the conveyor belt; see Figure 2 , which is the flow chart of rice conveying and single grain shedding in this step.

[0066] The rice to be sorted is spread on a vibrating conveyor belt made of food-grade material and has a fine grid texture on the surface. The texture depth and spacing are optimized according to the characteristics of the rice variety. The surface of the conveyor belt must maintain a certain degree of roughness to maintain friction with the rice, prevent the rice from relative slippage or rolling during transportation, and ensure that the rice can move synchronously with the conveyor belt; the surface parameters, effective working width and length of the conveyor belt are configured and replaced according to production capacity requirements to meet the production capacity requirements of sorting rice with different color grains.

[0067] A vibration motor is installed under the conveyor belt; a servo motor is selected as the vibration motor, and the vibration motor applies controllable vibration to the conveyor belt. The controllable vibration includes adjustable vibration frequency and vibration amplitude. By applying vibration, the rice grains on the conveyor belt are evenly dispersed.

[0068] Applied vibration frequency It is very important to achieve uniform dispersion of rice. It is necessary to consider factors such as rice variety, particle size distribution and required sorting capacity. The vibration frequency The calculation formula is: ;in, is the equivalent stiffness coefficient of the conveyor belt, which is related to parameters such as the material, size and tension of the conveyor belt; is the acceleration due to gravity; is the amplitude of the conveyor belt; is the mass per unit length of the conveyor belt; is the mass per unit length of the rice sample on the conveyor belt.

[0069] For example, for a typical rice variety, when the belt's equivalent stiffness coefficient is 750 N / mm, the amplitude can be set to 6 mm, the belt's mass per unit length is 2.5 kg / m, and the rice's mass per unit length is 0.75 kg / m, the calculated vibration frequency is approximately 25 Hz. By online monitoring of rice feeding and combining visual feedback, the vibration parameters are dynamically optimized to achieve optimal rice dispersion.

[0070] After the rice is fully dispersed by vibration, it is driven by the conveyor belt at a constant speed. Stable movement, the conveyor belt speed should be set according to the particle size, shape characteristics and vibration state of different rice varieties, so as to fully disperse the rice and avoid rice spilling due to excessive speed; conveyor belt speed The calculation formula is: ;in, is the distance that rice moves on the conveyor belt, that is, the effective length of the conveyor belt; The time required for rice to pass through the conveyor belt; is the frequency of the vibration motor; is the number of vibrations the rice receives on the conveyor belt.

[0071] For example, when the effective length of the conveyor belt is 2000 mm and the vibration frequency is 25 Hz, and the rice is expected to be vibrated 100 times during the conveyance process, it is calculated that the conveyor belt speed should be set to 500 mm / s. At this time, the time for the rice to pass through the conveyor belt is 4 seconds, which can obtain a sufficient vibration dispersion effect.

[0072] As the rice reaches the end of the conveyor belt, a guide rail system separates the individual grains and guides them downward. The rails are made of self-lubricating engineering plastic, ensuring minimal sliding resistance for the rice. The rails feature a trapezoidal cross-section, wide at the top and narrow at the bottom, helping to guide the rice toward the center. A gradient array of wedge-shaped grooves is machined at the rail entrance, gradually increasing in width from entrance to exit. This effectively diverts and guides the rice, preventing multiple grains from entering the same channel and causing blockage.

[0073] This step, through the synergistic effect of the conveyor belt's vibration dispersion and the guide rail's directional separation, efficiently transforms the rice from a disordered accumulation to an orderly, single-grain falling state. Operating parameters can be flexibly adjusted based on the physical properties of different rice varieties, demonstrating excellent adaptability and versatility. The vibration dispersion mechanism effectively breaks up clumping between rice grains, while the guide rail separation ensures the grains' individual alignment and controllable trajectory, creating ideal material flow conditions for subsequent high-speed image acquisition and intelligent recognition.

[0074] Step 2: Fit the rice falling trajectory into a trajectory surface and divide the trajectory surface into a collection area, a recognition area, and a separation area. Use an industrial camera to continuously capture images of falling rice in the collection area; see Figure 3 , which is the flow chart of rice falling area division and image acquisition in this step.

[0075] Mathematical modeling was performed on the falling trajectory of rice after leaving the conveyor belt. After leaving the conveyor belt, the rice moves in a parabolic motion under the action of gravity. Considering that there is only a slight difference in the spatial distribution of the conveyor rail outlet and the initial velocity of the rice grains, the motion trajectories of all rice grains form a regular parabola in three-dimensional space. A spatial coordinate system was established with the center of the guide rail outlet as the origin, the horizontal motion direction as the x-axis, the vertical downward as the z-axis, and the horizontal direction as the y-axis. The motion trajectory equation of a single rice grain is: , ,in, 、 are the coordinates of the rice grain in the horizontal and vertical directions at time t, is the initial velocity of the rice grain when it leaves the guide rail, which is equivalent to the conveyor belt speed. is the angle between the initial velocity and the horizontal plane, is the acceleration due to gravity, By performing statistical analysis on multiple trajectories, the trajectory envelope is fitted into a quadratic surface. ,in to is the fitting coefficient, which can be solved by the least square method. x, y, and z are the spatial coordinates of the rice during its fall, with the starting point of the fall as the origin.

[0076] Based on the fitted trajectory surface, the space is divided into three functional areas along the falling direction, including the collection area, the recognition area, and the separation area. The collection area is set at the initial falling section after the rice grains leave the guide rail. In this area, the distance between the rice grains is moderate and the movement is stable, which is conducive to obtaining clear individual images. The starting position height of the collection area is and end position height Determined based on the camera's depth of field and the speed of rice grain movement, satisfying the relationship ,in, is the effective depth of field of the industrial camera, The recognition area is below the acquisition area, which reserves a sufficient time window for image processing and feature recognition. ,in is the average falling speed of rice grains in this area, is the execution time of the image processing and recognition algorithms. The separation zone is located below the recognition zone and at the end of the falling trajectory. This is the area where the physical separation action is performed. The location and length of the separation zone are determined by the response time and range of the airflow nozzle.

[0077] The collection area uses a high-speed linear array CCD camera to collect images of falling rice. The high-speed linear array CCD camera is installed in the normal direction of the curved section of the collection area; the camera optical axis maintains a fixed distance from the curved surface of the rice grain movement trajectory. , this distance is determined by depth of field calculation ,in, is the focal length of the camera lens, is the aperture number, is the diameter of the circle of confusion, The camera uses continuous scanning mode, and the line frequency is Matches the falling speed of rice grains, satisfying ,in, is the vertical falling velocity of rice grains in the collection area, For example, when the falling speed of rice grains is 2.5 m / s and the spatial resolution is required to reach 0.1 mm, the camera line frequency needs to be set to 25 kHz.

[0078] To ensure the integrity and continuity of image acquisition, a synchronous trigger mechanism is configured in the acquisition area. The laser sensor detects whether rice enters the acquisition area. When a rice grain is detected entering the acquisition area, a trigger signal is generated to start the camera's image acquisition sequence. The sensor array is arranged in a matrix with a row spacing of and column spacing satisfy , ,in, and The minimum length and width of rice grains are set respectively. This dense arrangement ensures that every grain of rice can be reliably detected without missing any grains.

[0079] The high-speed linear array CCD camera uses a high-frequency flash LED array for lighting, which is synchronized with the camera exposure. The LED array is distributed in a ring around the camera lens, providing uniform diffuse lighting to eliminate specular reflections and shadows on the rice grain surface. The lighting intensity is dynamically adjusted through PWM to maintain the same lighting conditions during the separation of the same batch of different-colored rice, ensuring that the collected rice images have stable brightness and contrast. The flash frequency is strictly synchronized with the camera line frequency, and the pulse width is 1 / 4. , to avoid motion blur.

[0080] This step transforms the continuous falling rice grains into discrete functional zones through precise trajectory modeling and spatial segmentation. The parameters of each zone are optimized for its specific function. The rational placement of the acquisition zones ensures image quality, the recognition zones reserve ample processing time, and the precise positioning of the separation zones enhances sorting accuracy. The combination of a high-speed camera and synchronized lighting enables clear imaging of rapidly moving rice grains, providing high-quality raw data for subsequent image processing and feature recognition. This step effectively addresses the challenge of accurately capturing dynamic targets and lays a solid foundation for high-precision identification and classification of heterochromatic grains.

[0081] Step 3: Based on the color difference enhancement model of the generative adversarial network, the local contrast of the collected falling rice image is adjusted, and the dynamic range of the collected falling rice image is adjusted in the RGB and HSI dual domains; see Figure 4 , which is the flow chart of rice image chromatic aberration enhancement and dynamic range stretching in this step.

[0082] In this step, a specialized generative adversarial network (GAN) color difference enhancement model is constructed to address the extremely subtle color differences between slightly yellow rice, light yellow rice, and normal rice. This color difference enhancement model utilizes an improved conditional generative adversarial network architecture, with the generator network consisting of an encoder and a decoder. The encoder layer of the generator network uses a progressive convolutional layer structure to extract multi-scale features of the rice image layer by layer. Each convolutional layer is followed by a batch normalization layer and a Reluctant Unit (ReLU) activation function. The decoder layer reconstructs the enhanced image through deconvolution operations, introducing skip connections at each layer to preserve the details of the original image. The discriminator network uses a PatchGAN structure to divide the input image into multiple local regions for authenticity discrimination, improving sensitivity to local color difference features.

[0083] Loss function of the generator network Designed for ,in, To counter the loss, it is used to deceive the discriminator; It is a color enhancement loss, specifically optimized for subtle color differences; To maintain the loss of edge, to prevent the rice grain outline from being distorted during the enhancement process; 、 、 is the weight coefficient, which is dynamically adjusted according to the training process; color enhancement loss It is defined by calculating the color difference of the image before and after enhancement in Lab color space. ,in, Image position The weight factor at Image position This design enables the model to focus on enhancing areas with smaller color differences while maintaining the naturalness of areas with obvious color differences.

[0084] The rice image processed by the color difference enhancement model is subjected to local contrast amplification, and the local contrast amplification adopts the adaptive histogram equalization technology. Different from the traditional method, this application designs a dynamic block strategy according to the color distribution characteristics of the rice grain image; each rice grain image is divided into The size of each sub-block is determined adaptively according to the size of the rice grains, satisfying ,in, is the sub-block area, is the projected area of ​​the rice grain, is the block coefficient. Calculate the local contrast enhancement function for each sub-block ,in, is the original pixel value, is the sub-block mean, and is the linear transformation parameter, is the nonlinear enhancement index.

[0085] The dual-domain color shift stretching technology is used to process the collected rice images in both RGB and HSI color spaces. In the RGB domain, each color channel is independently stretched in dynamic range, and the stretching function is ,in, is the original red channel value, is the minimum value of the channel, is the stretch coefficient, The green and blue channels in the RGB domain are processed similarly, but the stretching coefficients are set differently based on the color difference contribution of each channel.

[0086] In the HSI domain, the hue H and saturation S components are enhanced, while the brightness I component remains basically unchanged. Hue enhancement uses piecewise linear mapping ,in, For enhanced color, It is a hue shift function. When the heterochromatic rice is set to slightly yellowish rice, the hue enhancement has a larger gain in the slightly yellowish hue range, while other hue ranges maintain a smaller shift to avoid color distortion.

[0087] This step can automatically learn and identify the unique color pattern of the off-color grain rice through the deep learning capability of the generative adversarial network, realize targeted enhancement, and use the local contrast amplification technology to improve the distinguishability of the subtle color difference in the image, so that the color difference that is difficult to detect with the naked eye becomes clear and visible. The double-domain color deviation stretching fully utilizes the advantages of different color spaces, enhances the color difference while maintaining the naturalness and authenticity of the image. The color difference between the slightly yellow rice and the normal rice is amplified by several times in the whole color difference enhancement process, which significantly improves the detection capability of the off-color grain rice with slight color difference.

[0088] Step 4: Establish a standard color library of rice, use a dynamic reference color spectrum two-way recognition algorithm and a relative color difference mapping strategy to identify off-color grain rice, and label in the identification area, while constructing an off-color grain feature reverse index mechanism for two-way identification of off-color grain rice and normal rice; see Figure 5 , the flowchart of off-color grain rice identification and labeling for this step;

[0089] A standard color library of rice is established as the identification reference, normal rice samples are collected, and accurate color parameters are obtained through a spectrophotometer and a standard light source box; the color feature vector of each sample is extracted in the CIELab color space , wherein is the lightness value, and are the color coordinates, is the chroma value, is the hue angle. The color data is divided into several standard color intervals through cluster analysis, and the center value of each interval is used as the standard color reference of this category. The color library can use a hierarchical storage structure, and multiple indexes can be established according to rice varieties, production areas, and seasons, etc. to facilitate fast retrieval and dynamic updating.

[0090] A dynamic reference color spectrum two-way recognition algorithm is constructed, the core of which is to establish an adaptive color comparison mechanism; the dynamic reference color spectrum two-way recognition algorithm first extracts the color features of the enhanced rice grain image, calculates the color difference distance between the standard color library and each reference point , wherein , and represent the lightness value and color coordinates of the rice grain to be tested, , and represent the lightness value and color coordinates reference of the normal rice. A dynamic threshold function is constructed according to the color difference distribution , wherein is the dynamic threshold, is the basic threshold, is the standard deviation of the current batch of color difference, is the mean color difference, and is the adaptive coefficient; the dynamic threshold mechanism can automatically adjust the discrimination criteria according to the overall chromaticity characteristics of different batches of rice, which can improve the adaptability of the algorithm.

[0091] Construct relative color difference mapping technology to achieve accurate identification by constructing a two-dimensional color difference distribution map; map the chromaticity coordinates of each grain of rice to - On the plane of the coordinate axis, calculate its offset vector relative to the normal rice cluster center ,in, Indicates the offset on the a* axis (representing the "red-green" direction in the Lab color space), It represents the offset on the b* axis (representing the "yellow-blue" direction in the Lab color space). Definition of relative color difference index ,in, is the modulus of the offset vector, is the normal rice clustering radius, is the angle between the offset vector and the yellowish feature direction, if the relative color difference index If the color exceeds the set threshold, the rice grain is determined to be a different color rice grain.

[0092] The rice with different colors is marked in real time in the recognition area, and the rice with different colors is marked with wireframes. Figure 6 This is a schematic diagram of the calibration results of heterochromatic rice grains. After calibrating the heterochromatic rice grains, laser positioning is used to obtain the spatial coordinates of each heterochromatic rice grain in the identification area, and the trajectory of the heterochromatic rice grains in the identification area is measured. Then, based on the trajectory of the heterochromatic rice grains in the identification area, the movement trajectory of the heterochromatic rice grains in the separation area is calculated, and the heterochromatic rice grains are separated in the separation area.

[0093] To achieve bidirectional discrimination between rice with off-color grains and normal rice, this application establishes a bidirectional discrimination mechanism for identifying rice color. The bidirectional discrimination mechanism is used to identify rice from both positive and negative perspectives. The positive discrimination is based on comparison with a standard colorimetric library to determine whether the rice grains deviate from the normal range, used to separate off-color grains from normal rice. The negative discrimination utilizes a reverse indexing mechanism for off-color grain characteristics to verify whether the rice grains determined to be off-color do in fact possess typical off-color characteristics, thereby achieving identification and separation of normal rice from off-color grains. This bidirectional discrimination mechanism can achieve reverse selection of normal rice from off-color grains, which helps reduce waste in the rice screening process.

[0094] Establish a reverse index mechanism for heterochromatic grain features; for each rice grain identified as heterochromatic, not only its chromaticity feature vector is recorded, but also a comprehensive feature vector including texture, shape and local color distribution is constructed. These features are generated a unique index key by a hash function , which is stored in the reverse index table. When the rice reverse selection is performed, the historical discrimination records are retrieved, and the feature matching degree ; the similarity of the current rice and the historical records is evaluated to identify whether the current rice is a color difference grain rice, wherein and are the current feature and historical feature vectors, respectively.

[0095] This step obtains reliable identification criteria by establishing a perfect standard colorimetric library, which can adapt to different varieties and batches of rice. The adaptive characteristics of the dynamic reference color spectrum algorithm ensure stable identification performance when the environmental conditions change. The relative color difference mapping technology converts abstract color differences into intuitive geometric relationships, improving the accuracy and interpretability of discrimination. The color difference grain feature reverse index mechanism can realize the reverse selection of rice; through multi-dimensional and multi-level identification strategies, this step realizes accurate discrimination of color difference grains such as slightly yellow rice, providing accurate and reliable decision-making basis for subsequent physical separation.

[0096] Step 5: According to the calibration results, the coordinates of the color difference grain rice in the separation zone are calculated, and the opening time of the air jet nozzle is controlled to blow the color difference grain rice into the color difference grain container; refer to Figure 7 , which is the air flow separation flow chart of the color difference grain rice in this step.

[0097] Establishing a trajectory prediction equation for color difference grain rice , wherein and are the predicted coordinates of the rice in the separation zone, and are the calibration coordinates of the identification zone, and are the velocity components of the rice in the identification zone, is the identification time, is the acceleration of gravity, and are air resistance coefficients. By measuring the motion parameters of the rice in the identification zone in real time with a high-speed linear array CCD camera, and dynamically correcting the resistance coefficients based on the mass and shape characteristics of the rice, a millimeter-level position prediction accuracy is achieved.

[0098] The air jet nozzles are arranged in an array and uniformly distributed in the lateral direction in the separation zone. Each nozzle is equipped with a high-speed electromagnetic valve with a response time of less than 5 milliseconds, which can be accurately opened and closed in an instant when the rice passes through.

[0099] The effective range of the air jet nozzle is determined by fluid dynamics simulation, and the air flow field distribution function is established, wherein For spatial points The air flow pressure at is the nozzle outlet pressure, is the nozzle position coordinate, 、 、 is the airflow diffusion parameter. Based on the predicted trajectory of the rice grain and the airflow field distribution, the optimal nozzle combination is automatically selected to ensure sufficient lateral thrust is applied to the target rice grain while minimizing the impact on the adjacent normal rice.

[0100] The airflow nozzle control strategy adopts a three-level control mechanism of prediction-trigger-feedback; in the prediction stage, the optimal triggering time is calculated based on the trajectory of the rice grains. ,in, The nozzle action height is the triggering stage, where the laser sensor array detects the actual arrival position of the rice grains. When the deviation between the detection signal and the predicted timing is less than the set threshold, the corresponding nozzle is immediately activated. The feedback stage verifies the separation effect through the downstream sensor and dynamically adjusts the injection pressure and duration. For example, when the different-colored rice grains enter the separation zone at a speed of 2.8m / s, it is predicted that they will reach the nozzle action position in 15 milliseconds. When the rice grains are about to arrive, the corresponding nozzle is activated 2 milliseconds before they are about to arrive. A lateral airflow of 0.25 MPa is applied at the moment the rice grains arrive, which lasts for 8 milliseconds and generates an impulse of about 15 N·ms, deflecting the rice grains to the container of different-color grains.

[0101] This step achieves accurate physical separation of off-color rice grains through the synergistic effect of precise trajectory prediction and high-speed airflow control. A prediction model that accounts for air resistance significantly improves position prediction accuracy, laying the foundation for precise separation. The array nozzle system's rapid response and precise control ensure that force is applied only to the target off-color rice grains, avoiding impact on normal rice. The adaptive airflow parameter optimization strategy can handle rice grains with various physical properties, improving the stability and reliability of separation. This separation process is efficient and precise, reliably removing off-color rice grains while minimizing the impact on normal rice, significantly improving the automation level and sorting accuracy of rice quality control.

[0102] Step 6: After separating the current batch of off-color rice, return the rice in the off-color rice container to the conveyor belt entrance, readjust the vibration parameters applied by the vibration motor to the conveyor belt and adjust the conveying speed of the conveyor belt, and then use the off-color rice feature reverse indexing mechanism to identify the normal rice and separate it with airflow to separate the normal rice from the off-color rice; see Figure 8 , which is the normal rice reverse selection flow chart of this step.

[0103] When selecting normal rice from rice with different colored grains, the number of separated rice with different colored grains (containing mixed normal rice) is significantly reduced compared to the number of rice without separation. Therefore, by increasing the time interval between the falling grains of different colored rice, normal rice can be selected more accurately to avoid the normal rice and the different colored grains from being mixed again.

[0104] The rice collected in the container of heterochromatic grains is returned to the entrance of the vibrating conveyor belt. In order to better separate the normal rice mixed with the heterochromatic grains, the vibration parameters of the conveyor belt need to be optimized and adjusted. First, the vibration frequency of the conveyor belt is reduced and the vibration amplitude is increased to make the rice grains more fully dispersed and jump on the conveyor belt, and the spacing between the rice grains on the conveyor belt is increased. The optimized vibration frequency The calculation formula is: ;in, is the equivalent stiffness coefficient of the optimized conveyor belt, The vibration amplitude of the conveyor belt after the increase is: is the acceleration due to gravity, is the mass per unit length of the conveyor belt, is the mass per unit length of the returned heterochromatic particles on the conveyor belt; by reducing and increasing the vibration amplitude can effectively reduce the vibration frequency and extend the movement time of rice grains on the conveyor belt.

[0105] After reducing the vibration frequency and increasing the vibration amplitude of the conveyor belt, the running speed of the conveyor belt is also reduced, further prolonging the retention time of rice grains on the conveyor belt, so that they can be more fully dispersed and jumped; the optimized conveyor belt speed The calculation formula is: ;in, is the effective length of the conveyor belt, is the time required for rice grains to pass through the conveyor belt, is the optimized vibration frequency, is the number of vibrations that rice grains experience on the conveyor belt. , which can make the rice grains obtain more complete vibration dispersion effect on the conveyor belt.

[0106] After optimization and adjustment, the returned rice with different colored grains achieved a larger spacing of dispersion and jumping movement on the conveyor belt, creating favorable conditions for the subsequent counter-selection process to accurately identify and separate normal rice.

[0107] When the rice grains pass through the recognition area again, the reverse index mechanism of the different-color grain features is activated for reverse selection, and the comprehensive feature vector is extracted for each grain of rice. , through the hash function Generate a unique index key, It is the unique index key of the current rice grain, and is used for quick matching in the reverse index table to identify whether the rice grain is normal rice. Unlike the forward discrimination during the initial sorting, the goal of the reverse selection is to identify normal rice.

[0108] Define feature similarity function ,in, is the feature vector of the current rice grain, is the normal rice feature vector stored in the reverse index table. Exceeds the preset similarity threshold When , the current rice grain is marked as normal rice and is used as the object of current airflow separation.

[0109] During the reverse rice selection process, in the separation zone of the rice falling surface, the airflow nozzle applies airflow force to the target marked as normal rice, deflecting it into the normal rice collection container; the control strategy of the airflow parameters is the same as that in step 5, and can be optimized in real time according to the physical properties and movement state of the rice grains to ensure accurate separation.

[0110] This step expands the falling distance of heterochromatic rice grains by optimizing and adjusting the parameters of the vibration conveyor belt, providing more favorable conditions for the accurate identification of reverse-selected normal rice; the application of hash index and feature similarity matching algorithm realizes the rapid screening of normal rice mixed in heterochromatic grains, greatly improving the recognition efficiency and accuracy. The perfect combination with the airflow separation technology ensures that normal rice can be reliably separated from heterochromatic grains, minimizing the waste of rice grains. The technical solution of this step forms a set of efficient, accurate and comprehensive screening solutions for positive and reverse selection of heterochromatic rice grains through the optimized combination of vibration dispersion, reverse index identification and airflow separation.

[0111] The technical solution for this step also makes full use of the big data resources accumulated from the initial color sorting. Combined with innovative algorithm design and sophisticated process control, it ensures the quality of the finished product while improving the rice output rate, realizing closed-loop optimization and intelligent upgrading of the color sorting process, and providing new ideas and methods for the high-quality development of the rice processing industry.

[0112] Example 2

[0113] Reference Figure 9 , which is the second embodiment of the present application, provides a dynamic grading and counter-selection control system for heterochromatic rice grains.

[0114] The system includes: a vibration transmission module, a falling partition module, an image enhancement module, a rice grain recognition module, a pneumatic separation module and a reverse selection module;

[0115] The vibration conveyor module applies vibration to the rice on the conveyor belt through a vibration motor, and a guide rail is provided at the end of the conveyor belt to allow single rice grains to fall off the conveyor belt;

[0116] The falling rice partitioning module fits the rice falling trajectory into a trajectory surface and divides the falling rice into a collection area, an identification area and a separation area according to the trajectory surface, and continuously collects falling rice images in the collection area using an industrial camera;

[0117] The image enhancement module performs local contrast adjustment on the collected falling rice image based on the color difference enhancement model of the generative adversarial network, and performs dynamic range adjustment on the collected falling rice image in the RGB and HSI dual domains;

[0118] The rice grain recognition module establishes a rice standard colorimetric library, utilizes a dynamic reference color spectrum bidirectional recognition algorithm and a relative color difference mapping strategy to discriminate heterochromatic rice, and performs calibration within the recognition area. It also constructs a heterochromatic grain feature reverse indexing mechanism to perform bidirectional identification of heterochromatic rice and normal rice.

[0119] The pneumatic separation module calculates the coordinates of the heterochromatic rice grains in the separation zone according to the calibration results, and controls the opening timing of the air flow nozzle to blow the heterochromatic rice grains into the heterochromatic rice container;

[0120] After separating the current batch of rice with different-colored grains, the reverse selection module returns the rice in the container with different-colored grains to the entrance of the conveyor belt, readjusts the vibration parameters applied by the vibration motor to the conveyor belt and adjusts the conveying speed of the conveyor belt, and then uses the reverse indexing mechanism of the different-colored grain characteristics to identify the normal rice and separate it with airflow, thereby separating the normal rice from the rice with different-colored grains.

[0121] In the embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0122] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose of this application and the scope of protection of the claims, which are all within the protection of this application.

Claims

1. A dynamic classification method for controlling the reverse selection of heterochromatic rice grains, characterized in that: include: The rice on the conveyor belt is vibrated by a vibration motor, and a guide rail is set at the end of the conveyor belt to make the single rice grains fall off the conveyor belt; The rice falling trajectory is fitted into a trajectory surface and divided into a collection area, a recognition area and a separation area according to the trajectory surface. An industrial camera is used to continuously capture images of falling rice in the collection area. The color difference enhancement model based on the generative adversarial network is used to adjust the local contrast of the collected falling rice image, and the dynamic range of the collected falling rice image is adjusted in the RGB and HSI dual domains; A rice standard colorimetric library was established, and a dynamic reference chromatogram bidirectional recognition algorithm and relative color difference mapping strategy were used to identify off-color rice grains. Calibration was performed within the recognition area, and a reverse indexing mechanism for off-color grain features was constructed to perform bidirectional identification between off-color rice grains and normal rice. The movement trajectory of the heterochromatic rice grains in the separation zone is calculated based on the calibration results, and the opening timing of the air flow nozzle is controlled to blow the heterochromatic rice grains into the heterochromatic rice grain container; After separating the current batch of rice with different colored grains, the rice in the container of different colored grains is returned to the entrance of the conveyor belt, and the vibration parameters applied by the vibration motor to the conveyor belt and the conveying speed of the conveyor belt are readjusted. The normal rice is then identified and separated by airflow using the reverse indexing mechanism of the different colored grain characteristics to separate the normal rice from the rice with different colored grains.

2. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 1, wherein: The rice to be sorted is conveyed onto a vibrating conveyor belt, and the conveyor belt is vibrated to increase the distance between the rice grains on the conveyor belt; A servo vibration motor is installed below the conveyor belt, and the servo vibration motor applies controllable vibration to the conveyor belt, and the controllable vibration includes adjustable vibration frequency and vibration amplitude. When the rice moves to the end of the conveyor belt, the guide rail device realizes the separation of the rice into single grains. The surface roughness, effective working width and length of the conveyor belt are configured according to production requirements.

3. A dynamic classification method for controlling the reverse selection of heterochromatic rice grains according to claim 2, characterized in that: The motion trajectories of all rice grains falling from the conveyor belt are fitted in three-dimensional space to form a trajectory surface. Based on the fitted trajectory surface, the trajectory surface is divided into multiple functional areas along the falling direction; The functional areas include a collection area, an identification area and a separation area; the collection area is arranged in the initial falling section where the rice grains leave the conveyor belt guide rail.

4. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 3, wherein: A high-speed linear array CCD camera is used in the collection area to collect images of the falling rice during the falling process. The high-speed linear array CCD camera is installed in the normal direction of the section of the trajectory surface of the collection area, and the optical axis of the camera maintains a fixed distance from the rice grain motion trajectory surface; The high-speed linear array CCD camera uses a high-frequency flash LED array for lighting during shooting, which is synchronized with the camera exposure. The LED array is distributed in a ring around the lens of the high-speed linear array CCD camera. The lighting intensity of the LED array is dynamically adjusted through pulse width modulation, and the flash frequency is synchronized with the camera line frequency.

5. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 4, wherein: Constructing a color difference enhancement model, wherein the color difference enhancement model is constructed using an improved conditional generative adversarial network architecture, wherein the generative adversarial network architecture includes a generator network and a discriminator network; The generator network includes an encoder and a decoder. The encoder adopts a progressive convolutional layer structure. The decoder reconstructs the enhanced falling rice image through deconvolution operations and introduces skip connections in each layer. The discriminator network adopts the PatchGAN structure to divide the input falling rice image into multiple local areas for authenticity discrimination.

6. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 5, wherein: performing local contrast amplification on the falling rice image processed by the chromatic aberration enhancement model, wherein the local contrast amplification adopts an adaptive histogram equalization technique, designs a dynamic block strategy according to the color distribution characteristics of the falling rice image, divides each falling rice image into sub-blocks, and the area size of the sub-blocks is adaptively determined according to the size of the rice grains; The falling rice image is subjected to local contrast amplification and then processed simultaneously in RGB and HSI color spaces using a dual-domain color shift stretching technique. In the RGB domain, dynamic range stretching is performed independently on each color channel, and in the HSI domain, hue and saturation components are enhanced.

7. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 6, wherein: A rice standard chromaticity library is constructed as a rice identification benchmark. The standard chromaticity library obtains chromaticity parameters by collecting normal rice samples, extracts chromaticity feature vectors in the CIELab color space for each sample, and divides the chromaticity parameters into multiple standard chromaticity intervals by cluster analysis; A dynamic reference chromatogram bidirectional recognition algorithm is constructed. The dynamic reference chromatogram bidirectional recognition algorithm extracts chromaticity feature vectors from the enhanced falling rice image, calculates the color difference distance between the chromaticity feature vector extracted from the falling rice image and each reference point in the standard chromaticity library, and constructs a dynamic threshold function based on the color difference distribution. The dynamic threshold function automatically adjusts the discrimination standard according to the overall chromaticity characteristics of different batches of rice; A relative color difference mapping strategy was constructed to identify normal rice and rice with different colored grains by constructing a two-dimensional color difference distribution map.

8. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 7, wherein: The heterochromatic rice grains are calibrated in real time in the identification area. After the heterochromatic rice grains are calibrated, the spatial coordinates of each heterochromatic rice grain in the identification area are obtained through laser positioning, and the trajectory of the heterochromatic rice grains in the identification area is measured. The movement trajectory of the heterochromatic rice grains in the separation area is calculated based on the trajectory of the heterochromatic rice grains in the identification area.

9. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 8, wherein: Establish a reverse indexing mechanism for heterochromatic grain features. For each rice grain identified as heterochromatic, record the chromaticity features of the rice and construct a comprehensive feature vector for the heterochromatic grain, including texture, shape, and local color distribution. The normal rice in the collected heterochromatic rice is reversely selected through the heterochromatic grain feature reverse index mechanism to separate the normal rice from the heterochromatic grain rice.

10. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 9, characterized in that: In the separation zone, the different-colored rice grains in the falling rice are blown out by air flow, and the air flow is blown out from the air flow nozzle corresponding to the separation zone; The airflow nozzles are arranged in an array, with multiple airflow nozzles evenly distributed laterally in the separation zone. Each airflow nozzle is opened and closed by a high-speed solenoid valve. The effective range of the airflow nozzle is determined by fluid dynamics simulation. The control strategy of the airflow nozzle adopts a three-level control mechanism of prediction-trigger-feedback. During the trigger stage, the actual arrival position of the rice grains is detected by a laser sensor array. When the deviation between the detection signal and the predicted timing is less than the set threshold, the corresponding nozzle is activated to blow the heterochromatic rice grains into the heterochromatic rice container.

11. The method for controlling the reverse selection of rice with different-colored grains by dynamic classification according to claim 10, characterized in that: The rice collected in the container of rice with different colored grains is fed back to the entrance of the vibrating conveyor belt, and the vibration parameters and operating speed of the conveyor belt are optimized and adjusted to increase the spacing between the rice grains on the conveyor belt; When the rice grains pass through the identification zone again, the reverse index mechanism for the characteristics of the heterochromatic grains is activated to identify normal rice. For each grain of rice, its comprehensive feature vector is extracted and matched and identified in the reverse index table; During the reverse selection process, normal rice is blown out through the air flow nozzle in the separation area of ​​the rice falling surface, so that the normal rice is deflected into the collection container during the falling process.

12. A dynamic classification and counter-selection control system for rice with different colored grains, which is used to implement the dynamic classification and counter-selection control method for rice with different colored grains according to any one of claims 1 to 11, characterized in that: include: Vibration transmission module, falling partition module, image enhancement module, rice grain recognition module, pneumatic separation module and anti-selection module; The vibration conveyor module applies vibration to the rice on the conveyor belt through a vibration motor, and a guide rail is provided at the end of the conveyor belt to allow single rice grains to fall off the conveyor belt; The falling rice partitioning module fits the rice falling trajectory into a trajectory surface and divides the falling rice into a collection area, an identification area and a separation area according to the trajectory surface, and continuously collects falling rice images in the collection area using an industrial camera; The image enhancement module performs local contrast adjustment on the collected falling rice image based on the color difference enhancement model of the generative adversarial network, and performs dynamic range adjustment on the collected falling rice image in the RGB and HSI dual domains; The rice grain recognition module establishes a rice standard colorimetric library, utilizes a dynamic reference color spectrum bidirectional recognition algorithm and a relative color difference mapping strategy to discriminate heterochromatic rice, and performs calibration within the recognition area. It also constructs a heterochromatic grain feature reverse indexing mechanism to perform bidirectional identification of heterochromatic rice and normal rice. The pneumatic separation module calculates the movement trajectory of the heterochromatic rice grains in the separation zone according to the calibration results, and controls the opening timing of the air flow nozzle to blow the heterochromatic rice grains into the heterochromatic rice container; After separating the current batch of rice with different-colored grains, the reverse selection module returns the rice in the container with different-colored grains to the entrance of the conveyor belt, readjusts the vibration parameters applied by the vibration motor to the conveyor belt and adjusts the conveying speed of the conveyor belt, and then uses the reverse indexing mechanism of the different-colored grain characteristics to identify the normal rice and separate it with airflow, thereby separating the normal rice from the rice with different-colored grains.

Images