Uniform light supplementing method and system for full-view imaging of conveying belt

Abstract

The invention relates to a uniform light supplementing method and system for full-view imaging of a conveying belt, and the method is realized based on all modules of the system, and comprises the following steps: obtaining a belt region image; and inputting the belt area image into a pre-training model, obtaining an identification result output by the pre-training model, generating an adjustment scheme based on the identification result and a preset light supplement rule, and outputting and completing belt area light supplement based on the adjustment scheme. According to the method, image shadow, glare and brightness difference caused by uneven illumination are eliminated, full-view illumination uniformity is ensured, the image is clear, the color is real, details are rich, and a high-quality basis is provided for subsequent analysis. Stable and consistent illumination conditions are provided for industrial detection, the false detection and omission ratio is reduced, and the product quality is ensured. The system adaptability is enhanced, the requirements of different industrial scenes are met, and uniform light supplementing can be achieved regardless of the target shape, material and surface characteristics.

Description

Technical Field

[0001] This application relates to the field of industrial automation technology, and particularly to a method and system for uniform supplementary lighting for full-field imaging of a conveyor belt. Background Art

[0002] With the rapid development of industrial automation and intelligent manufacturing, the quality inspection, dimension measurement, target recognition and other links in industrial production rely more and more on the vision system. However, in the actual industrial environment, the lighting conditions are often complex and changeable, making it difficult to meet the requirements of high-quality image acquisition. Therefore, it is necessary to supplement light with a conveyor belt light source to obtain clear and accurate images.

[0003] Traditional industrial light source supplementary lighting methods have many problems. For example, a light source with a fixed brightness cannot adapt to different environments and scenarios, easily resulting in overexposure or underexposure; manually adjusting the light source brightness requires manual intervention, cannot achieve automation, and depends on the operator's experience, resulting in unstable image quality; the method of adjusting based on ambient brightness feedback simply responds to changes in ambient brightness without considering the actual quality and content of the image; the method of adjusting based on image content may over-respond to specific image content.

[0004] In industrial vision inspection, the quality of the collected images directly affects the subsequent image recognition, analysis and processing results. Uneven supplementary lighting will cause shadows, highlights and dark areas in the images, affecting the extraction and judgment of object features, and reducing the detection accuracy and reliability. Therefore, the full-field imaging uniform supplementary lighting technology has become the key to improving the performance of industrial vision inspection systems. The conveyor belt environment has dim light and uneven local lighting. At the same time, the material characteristics of the belt surface cause different degrees of reflection at different lighting angles, especially when there are water stains or material dust attached, the reflection phenomenon is more serious. This will not only cause local over-brightness or over-darkness in the image, but also cover up the defects and detail information on the belt surface, increasing the difficulty of image processing and analysis. Summary of the Invention

[0005] To solve or partially solve the problems existing in the related technologies, this application provides a method and system for uniform supplementary lighting for full-field imaging of a conveyor belt, eliminating image shadows, glare and brightness differences caused by uneven lighting, ensuring uniform full-field lighting, making the images clear, with true colors and rich details, and providing a high-quality basis for subsequent analysis.

[0006] The first aspect of this application provides a method for uniform supplementary lighting for full-field imaging of a conveyor belt, including the following steps:

[0007] Obtain an image of the belt area;

[0008] Input the belt area image into the pre-trained model to obtain the recognition result output by the pre-trained model. Here, the pre-trained model is a comprehensive model including a light information extraction and analysis model and an object feature recognition model. The light analysis model is used to identify the position and range of the uneven light area in the belt area image, and the object feature recognition model is used to extract and classify the material and shape features of the objects on the belt;

[0009] Generate an adjustment plan based on the recognition result and the preset light filling rule, output it, and complete the light filling of the belt area based on the adjustment plan.

[0010] Among them, the adjustment plan includes:

[0011] Dynamically adjust the light source intensity, and change the irradiation angle of the light source array in real time according to the shape and position of the object;

[0012] Switch the light filling mode according to the detection requirements of objects with different materials, and change the optical characteristics through intelligent optical materials; determine the position and angle of the reflector, and construct a reflection system to evenly reflect the light to the target area;

[0013] In addition, calculate the adjusted value of the light filling parameter according to the uneven light area, shadow and highlight conditions to achieve uniform light filling.

[0014] Among them, the processing steps of the light information extraction and analysis model include:

[0015] Obtain the light information of the area based on the belt area image, and calculate the brightness difference value of each area of the image;

[0016] Compare the brightness difference value with the preset uniformity threshold to identify the underlit area and the overlit area;

[0017] Mark the position and range of the uneven light area. The uneven light area includes the underlit area, the overlit area, as well as the shadow and highlight areas.

[0018] Among them, the preset light filling rules include the light intensity adjustment rule, the light source angle adjustment rule, the light filling mode switching rule, the light uniform distribution rule and the real-time image analysis feedback rule.

[0019] Among them, the light intensity adjustment rule and the light source angle adjustment rule are: according to the recognition result of the uneven light area, automatically increase the output power of the corresponding light source in the underlit area, and reduce the corresponding light source intensity in the overlit area; use the constructed dynamically adjustable light source array, through motor drive and high-precision control circuit, according to the shape and position changes of the target object, change the light irradiation angle and intensity in real time.

[0020] Among them, the light filling mode switching rule includes:

[0021] When the material of the object is identified as metal and cracks need to be detected, start the fill light mode integrating a high-brightness LED and a gas discharge lamp in the ultraviolet band.

[0022] When it is recognized that the belt area is in a complex light environment, automatically adjust the light transmittance and color through an electrochromic filter to eliminate ambient light interference.

[0023] Among them, the light uniform distribution rules include:

[0024] When the object is a large workpiece, use multiple mirror groups to reflect the light emitted by the light source multiple times to cover the entire field of view; when the object is a large irregular mechanical part, automatically identify the part contour and adjust the light source array to achieve uniform fill light.

[0025] Among them, the real-time image analysis feedback rules include that when defects are found on the surface of the object packaging, identify the shadows generated by the wrinkles and adjust the fill light parameters to eliminate the influence.

[0026] Among them, the object feature recognition model is a convolutional neural network model, including:

[0027] The convolutional layer is used to perform convolutional operations on the belt area image to extract edge and texture features;

[0028] The pooling layer is used to reduce the resolution of the feature map;

[0029] The residual block deepens the network depth and solves the gradient disappearance problem to learn high-level features;

[0030] The fully connected layer maps the extracted features to different categories to complete the classification and recognition of the object material and shape.

[0031] The second aspect of this application provides a uniform fill light system for full-field imaging of a conveyor belt, which is applicable to the uniform fill light method for full-field imaging of a conveyor belt as shown in the first aspect, including:

[0032] The image acquisition module includes several image sensors installed in the belt area and is used to acquire belt area images;

[0033] The light analysis module is used to receive the belt area image, calculate the brightness difference of each area of the image, and output the adjustment instruction for the luminous intensity of different light sources in the light source array according to the preset uniformity threshold;

[0034] The object feature recognition module: is used to extract and classify the material and shape features of the object on the belt;

[0035] The light adjustment execution module: is used to receive the adjustment instruction for the luminous intensity, and through the motor drive and the high-precision control circuit, the brightness and irradiation angle of each light source in the dynamically adjustable light source array are changed in real time;

[0036] The fill light mode switching module is used to control the combined activation of different types of light sources and adjust the optical properties of lenses and filters made of intelligent optical materials;

[0037] The light reflection control module is used to control the positions and angles of the mirrors in the reflection system composed of multiple mirrors, and evenly reflect the light emitted by the light source to the target area;

[0038] The real-time image feedback analysis module is used to identify uneven illumination areas, shadows, and highlights, quickly calculate the fill light parameters to be adjusted according to the analysis results, and feedback them to the illumination adjustment execution module for real-time adjustment;

[0039] The distributed control module connects the light sources, optical components, and sensor devices in the fill light system into a network, and is responsible for coordinating the work of each device based on the main controller.

[0040] The technical solution provided by this application may include the following beneficial effects:

[0041] This application provides a method and system for uniform fill light for full-field imaging of a conveyor belt, aiming to solve the problem of industrial vision imaging, improve and ensure image quality. Eliminate image shadows, glare, and brightness differences caused by uneven illumination, ensure uniform illumination in the full field of view, make the image clear, with true colors and rich details, providing a high-quality basis for subsequent analysis. Provide stable and consistent illumination conditions for industrial inspection, avoid shadows caused by uneven illumination, so as to clearly present the details of the product, such as tiny scratches on the surface of components, solder joint quality, etc., greatly improving the clarity and accuracy of imaging, and effectively reducing the misjudgment rate. Meet the requirements of different industrial scenarios, and can achieve uniform fill light regardless of the shape, material, and surface characteristics of the target. It has a positive effect on industrial production efficiency, with a high success rate of imaging in one shot, reducing operations such as reshooting and inspection due to lighting problems, and improving the overall production efficiency. The adaptability to the working environment has been improved, providing stable and uniform illumination conditions in different industrial environments, such as darker factories or spaces with local obstructions, ensuring that the imaging effect is not overly interfered by ambient light, and enhancing the applicability and stability of the system.

[0042] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit this application. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] By describing the exemplary embodiments of this application in more detail in conjunction with the accompanying drawings, the above and other objects, features, and advantages of this application will become more obvious. Among them, in the exemplary embodiments of this application, the same reference numerals generally represent the same components.

[0044] Figure 1It is a schematic flowchart of the method shown in the embodiments of the present application. Detailed implementation manners

[0045] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0046] It should be understood that although the terms "first", "second", "third", etc. may be used in the present application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined by "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present application, the meaning of "a plurality" is two or more, unless otherwise specifically defined.

[0047] In the description of the present application, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present application.

[0048] Unless otherwise clearly defined and limited, the terms "mounted", "connected", "connected to", "fixed", etc. should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or integrated; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the internal communication of two elements or the interaction relationship between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0049] The technical solutions of the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0050] Embodiment 1:

[0051] A full - field imaging uniform light - supplementing system for a conveyor belt designs standardized interfaces, enabling the light - supplementing system to seamlessly access industrial automation production lines. Through industrial Ethernet or fieldbus, it realizes communication with devices such as controllers and robots on the production line.

[0052] It includes the following modules:

[0053] Image acquisition module: Composed of an image sensor, it is responsible for real - time acquisition of industrial scene images in the imaging area, converting optical signals into digital signals, and providing raw data for subsequent processing.

[0054] Lighting analysis module: Based on an adaptive illuminance adjustment algorithm, it receives the image data transmitted from the image acquisition module, calculates the brightness differences in each area of the image, and according to the preset uniformity threshold, outputs the instruction for adjusting the luminous intensity of different light sources in the light source array, ensuring that the full - field lighting uniformity is maintained above 80%.

[0055] Object feature recognition module: With a deep - learning model (such as the ResNet - 50 convolutional neural network) as the core, it successively performs convolution, pooling, residual block processing, and fully - connected operations on the input image, extracts the low - level and high - level features of the object, so as to accurately identify the material and shape category of the object, providing a basis for switching the light - supplementing mode.

[0056] Lighting adjustment execution module: Receives the instruction for adjusting the luminous intensity of the light source given by the lighting analysis module, as well as the signal of the change in the shape and position of the target object (from relevant modules such as position detection), and through motor drive and high - precision control circuits, it can change the brightness and irradiation angle of each light source in the dynamically adjustable light source array in real - time, enhancing the light source power in the under - illuminated area and reducing the light source intensity in the over - illuminated area.

[0057] Light - supplementing mode switching module: Based on the recognition results of the object material, shape, etc. output by the object feature recognition module, it controls the combined activation of different types of light sources (such as high - brightness LEDs and special gas - discharge lamps), as well as the adjustment of the optical properties of lenses and filters made of intelligent optical materials, eliminating ambient light interference and meeting different detection requirements.

[0058] Light reflection control module: By computer - simulating the light propagation path, based on the law of light reflection, it accurately calculates and controls the position and angle of each mirror in the reflection system composed of multiple mirrors, evenly reflecting the light emitted by the light source to the target area, solving the problem of insufficient lighting in the edge area during the detection of large workpieces.

[0059] Real-time Image Feedback Analysis Module: Continuously analyze the acquired images in real time using deep learning algorithms, identify uneven illumination areas, shadows, and highlights, and quickly calculate the supplementary lighting parameters to be adjusted (such as light source brightness, angle, etc.) based on the analysis results, and feedback them to the lighting adjustment execution module for real-time adjustment.

[0060] Distributed Control Module: Connect devices such as light sources, optical components, and sensors in the supplementary lighting system into a network. The main controller is responsible for coordinating the work of each device. Each distributed node performs preliminary processing based on local sensor data, reducing the burden on the main controller, improving the overall response speed of the system, and ensuring the coordinated operation between modules.

[0061] Design the supplementary lighting system for dust prevention, waterproofing, and electromagnetic interference prevention. Use a sealed housing and special protective coatings to protect internal optical and electronic components. In an environment with strong electromagnetic interference, add a shielding layer and filtering circuit to ensure the stable operation of the system. Continuously monitor the light source temperature through a temperature sensor, automatically adjust the cooling system, ensure that the light source is within the optimal operating temperature range, and extend its service life.

[0062] Embodiment 2:

[0063] As Figure 1 shown, a method for uniform supplementary lighting of full-field imaging of a conveyor belt is applicable to the full-field imaging uniform supplementary lighting system of the conveyor belt as shown in Embodiment 1, and includes the following steps:

[0064] S1. Obtain an image of the belt area.

[0065] Full-field imaging aims to obtain complete surface information of the conveyor belt, including the width direction and length direction of the belt. By adopting special optical systems, image stitching techniques, or multi-camera collaborative working methods, the problem of limited field of view of a single lens is overcome, and high-resolution imaging of a large-area belt surface is achieved.

[0066] The line array camera obtains image information by scanning line by line. In theory, it can achieve infinite-length image acquisition. Install the line array camera perpendicular to the running direction of the belt, and cooperate with the uniformly moving belt. The line array camera can scan the belt surface line by line to form a complete image of the belt surface. This method can achieve high-resolution full-field imaging, but has high requirements for the stability of the belt running speed. To ensure the imaging quality, it is necessary to accurately control the synchronization relationship between the belt speed and the scanning frequency of the line array camera.

[0067] During the image acquisition process, according to the requirements of the full-field imaging method, the parameters of the camera are set, such as exposure time, frame rate, resolution, etc. For line-scan camera imaging, the synchronous operation of the belt speed and the camera scanning frequency needs to be accurately controlled. The acquired images are first preprocessed, including operations such as denoising and grayscale transformation, to improve the image quality. Then, according to the different imaging methods, corresponding post-processing is carried out. For fisheye lens images, distortion correction is performed; for multi-camera mosaicked images, image mosaicking and fusion processing are carried out. Finally, the processed images are analyzed using target detection algorithms to identify information such as the running state, defects, and foreign objects of the belt.

[0068] S2. Input the belt area image into the pre-trained model to obtain the recognition result output by the pre-trained model.

[0069] The pre-trained model is a comprehensive model including a lighting information extraction and analysis model and an object feature recognition model. The lighting analysis model is used to identify the position and range of uneven lighting areas in the belt area image, and the object feature recognition model is used to extract and classify the material and shape features of the objects on the belt.

[0070] The lighting information extraction and analysis model uses an image sensor to collect the lighting information of the imaging area in real time, converts it into a digital signal and inputs it into the lighting analysis module. The brightness difference of each area of the image is calculated by the lighting analysis module, and the calculation formula is:

[0071]

[0072] where is the average image brightness, and n is the number of image areas.

[0073] By comparing the brightness difference D with the preset uniformity threshold T, the position and range of the uneven lighting area are identified. For the underlit area and the overlit area (L i > L), they are marked as uneven lighting areas, and their positions and ranges are recorded.

[0074] The object feature recognition model combines a convolutional neural network (CNN) to perform feature recognition and analysis on objects of different materials and shapes. A convolutional neural network model represented by ResNet-50 is built, which includes multiple convolutional layers, pooling layers, and fully connected layers.

[0075] The input belt image undergoes a series of convolution operations to extract low-level features (such as edges, textures, etc.) in the image. Then, through the pooling layer, the resolution of the feature map is reduced to reduce the computational amount. Next, it passes through multiple residual blocks to deepen the network depth and further learn more advanced features. Finally, through the fully connected layer, the features are mapped to different categories to achieve the classification of the object's material and shape. The residual block is between the convolutional layer and the fully connected layer and after the pooling layer. In the object feature recognition model process, the image first extracts low-level features through the convolutional layer, then the pooling layer reduces the resolution of the feature map to reduce the computational amount, and then the residual block plays a role. The residual block consists of multiple convolutional layers. Its unique feature is the introduction of cross-layer connections, allowing the original input features to directly skip some convolutional layers and be added to the features after convolutional processing. This structure effectively solves the problem of gradient disappearance during the training of deep neural networks, helps the network better learn advanced features, and lays a foundation for the subsequent fully connected layer to map the features to different categories to identify the object's material and shape.

[0076] When training this model, a large industrial image dataset containing objects of different materials and shapes is used, and the cross-entropy loss function is used for optimization. The model parameters are updated through optimization algorithms such as Stochastic Gradient Descent (SGD).

[0077] The final recognition results include:

[0078] Recognition result of uneven illumination area: Identify the positions and ranges of areas with insufficient illumination, excessive illumination, shadows, highlights, etc. in the image.

[0079] Object feature recognition result: Accurately identify the material and shape categories of the target object.

[0080] S3. Generate an adjustment plan based on the recognition results and preset fill light rules, output it, and complete the fill light of the belt area based on the adjustment plan.

[0081] The core principle of fill light is to artificially increase the scene illumination intensity and improve the illumination distribution to meet the imaging requirements of the vision system. Appropriate fill light can improve the image contrast, reduce noise interference, enhance the feature representation of the target object in the image, and thus improve the detection ability of the vision system for the belt running state. Multiple light sources are set according to the requirements, including:

[0082] White LED light source: White LEDs have the advantages of high luminous efficiency, long lifespan, fast response speed, etc. In the vision system of coal mine transportation belts, strip or ring-shaped white LED light sources can be used to fill light the belt surface. Strip light sources are suitable for uniform illumination along the length of the belt, while ring light sources can be used for key illumination of specific detection areas (such as belt joints) to reduce shadow generation.

[0083] Color LED light source: According to different detection requirements, an LED light source of a specific color can be selected. For example, when detecting foreign objects on the surface of coal and conveyor belts, by taking advantage of the reflection differences of certain foreign objects, coal, and conveyor belts for specific colors of light, an LED light source of the corresponding color is used for supplementary lighting, which can enhance the contrast between the foreign objects and the background and improve the detection accuracy.

[0084] Near-infrared LED light source: Near-infrared light has the characteristics of strong penetrability and is not easily interfered by ambient light. In underground coal mines, coal dust, fog, etc. will affect the propagation of visible light, but near-infrared light is less affected by them. The near-infrared LED light source can be used for supplementary lighting of conveyor belts in harsh environments. Cooperating with a camera with near-infrared light-sensitive ability, a clear image of the conveyor belt can be obtained. In addition, near-infrared light is harmless to the human body and will not interfere with the vision of coal miners.

[0085] Infrared laser light source: The infrared laser light source has the advantages of strong directivity and concentrated energy. After diffusing the infrared laser beam, uniform supplementary lighting of the conveyor belt can be carried out at a long distance. It is especially suitable for segmented supplementary lighting of long-distance transportation conveyor belts, which can ensure the uniformity of the light intensity within a large detection range.

[0086] The preset supplementary lighting rules are a series of pre-set criteria used to guide the adjustment of supplementary lighting according to the recognition results, mainly including the following aspects:

[0087] Light intensity adjustment rule: When the light information extraction and analysis model identifies an area with insufficient light, the power of the corresponding light source is increased proportionally. For example, when the light at the target edge is insufficient, the output power of the light source in this area is increased by a certain proportion (such as 1.2 times); for an area with excessive light, the light intensity is reduced by a proportion (such as 0.8 times) to balance the light.

[0088] Light source angle adjustment rule: Based on the shape and position changes of the target object, and based on the geometric relationship between the edge of the object and the relative position of the light source, the irradiation angle of the light source is calculated and adjusted through a specific formula (such as a trigonometric function formula) so that the light can irradiate each part of the target object more evenly and effectively.

[0089] Supplementary lighting mode switching rule: After the object feature recognition model identifies objects of different materials and shapes, the supplementary lighting mode is switched according to specific detection requirements. For example, when detecting micro-cracks on the surface of a metal, an integrated mode of a high-brightness LED and a gas discharge lamp in the ultraviolet band is enabled; for general detection in an environment with complex light, according to the comparison result between the ambient light intensity and the preset threshold, it is decided whether to enable the electrochromic filter to adjust the optical properties. For example, when the ambient light intensity exceeds the preset threshold, the electrochromic filter adjusts the light transmittance and color according to a specific formula.

[0090] Rule for uniform light distribution: For the inspection of large workpieces, according to the law of light reflection, the position and angle of the reflector are determined by computer simulation of the light propagation path, so that the reflection system composed of multiple reflectors can reflect the light of the light source multiple times, making the light evenly cover the entire field of view and solving the problem of insufficient illumination in the edge area.

[0091] Rule for real-time image analysis and feedback: When the deep learning algorithm analyzes the image in real time, for the identified uneven illumination areas, shadows and highlights, according to the preset parameters related to the defect type and area characteristics, the adjustment value of the supplementary light parameters is calculated. For example, when detecting the wrinkles and shadows on the food packaging, according to the area and average brightness of the shadow area, combined with the preset shadow adjustment parameters, the adjustment values of the light source brightness and angle are calculated.

[0092] Corresponding to the above preset supplementary light rules, an adjustment plan is obtained in combination with the recognition result, including:

[0093] Lighting adjustment: Dynamically adjust the light source intensity and change the irradiation angle of the light source array in real time according to the shape and position of the object.

[0094] According to the recognition result of the uneven illumination area, for the area with insufficient illumination, such as when the illumination of the target edge area is insufficient, the output power of the corresponding light source in this area is automatically increased. For the area with too strong illumination, the intensity of the corresponding light source is reduced.

[0095] Using the constructed dynamically adjustable light source array, through the motor drive and high-precision control circuit, according to the shape and position changes of the target object, the irradiation angle and intensity of the light source are changed in real time.

[0096] The design of the multi-component composite light source includes the integration of the hybrid spectrum light source and the dynamically adjustable light source array. The integration of the hybrid spectrum light source innovatively combines different types of light sources, such as the integration of high-brightness LEDs and special gas discharge lamps. The LED provides stable and energy-saving basic white light illumination, and the gas discharge lamp emits specific band light with high energy density for the detection requirements of specific materials, enhancing the extraction of target features. For example, when detecting the micro-cracks on the metal surface, the gas discharge lamp in the ultraviolet band is used in combination with the LED white light to make the cracks appear more clearly in the imaging.

[0097] The dynamically adjustable light source array constructs a dynamically adjustable light source array, and each light source unit has independent brightness and angle adjustment capabilities. Through the motor drive and high-precision control circuit, according to the shape and position changes of the target object, the irradiation angle and intensity of the light source are changed in real time. For example, when detecting large irregular mechanical parts, the system automatically identifies the part contour and adjusts the light source array to ensure that each part can obtain uniform and sufficient illumination.

[0098] Supplementary light mode switching: Switch the supplementary light mode according to the detection requirements of objects with different materials, and change the optical characteristics through intelligent optical materials.

[0099] For the detection of micro-cracks on the metal surface, when the material of the object is identified as metal and crack detection is required, the supplementary lighting mode integrated with a high-brightness LED and a gas discharge lamp in the ultraviolet band is activated. The LED provides stable and energy-saving basic white light illumination, and the gas discharge lamp emits ultraviolet band light with high energy density to enhance the extraction of crack features.

[0100] Lenses and filters made of electrochromic, liquid crystal and other intelligent optical materials change their optical properties in real time according to the ambient light changes and detection requirements. In the complex workshop environment with various light conditions, the electrochromic filter automatically adjusts the light transmittance and color to eliminate the interference of ambient light and ensure the imaging quality.

[0101] Adjustment of uniform light distribution: Design a reflection system composed of multiple reflectors. By accurately calculating the light propagation path, the reflectors are used to evenly reflect the light to the target area. For the detection of large workpieces, multiple mirror groups can reflect the light emitted by the light source multiple times to cover the entire field of view and solve the problem of insufficient illumination in the edge area. Utilizing the law of reflection of light that the angle of incidence is equal to the angle of reflection, the position and angle of the reflectors are determined by computer simulation of the light propagation path to ensure uniform light distribution.

[0102] Real-time image analysis and feedback adjustment: Use deep learning algorithms to perform real-time analysis on the collected images to identify uneven illumination areas, shadows and highlights in the images. The algorithm quickly calculates the supplementary lighting parameters that need to be adjusted, such as the brightness and angle of the light source, to achieve uniform supplementary lighting. When detecting surface defects of food packaging, the algorithm can quickly identify the shadows caused by wrinkles on the packaging, and based on the area and average brightness of the shadow area, combined with the preset shadow adjustment parameters, calculate the light source brightness adjustment value and angle adjustment value to eliminate the shadow.

[0103] Distributed control and collaborative adjustment: Build a distributed control system to connect devices such as light sources, optical components, and sensors in the supplementary lighting system into a network. The main controller coordinates the work of each device to achieve precise control of the supplementary lighting system. Each distributed node has a certain intelligent processing ability and can perform preliminary processing according to local sensor data to reduce the burden on the main controller and improve the system response speed. For example, the light source node preliminarily judges whether the light source needs to adjust its brightness according to the local light sensor data, and then uploads the result to the main controller. The main controller makes the final adjustment decision by integrating the information of each node.

[0104] Based on the width and running speed of the belt and the installation position of the imaging device, this method rationally designs the layout of the light source. For the case of using multiple light sources, it is necessary to ensure the uniform superposition of the light rays between the light sources to avoid the appearance of bright spots or shadows. At the same time, through experiments and simulation means, parameters such as the brightness, color, and irradiation angle of the light source are optimized. For example, when using a white LED strip light source, by adjusting the height and angle of the light source, the illumination intensity on the belt surface is evenly distributed, and the belt texture and coal particles in the image are clearly visible. Under different lighting environments and belt running states, the light source parameters can also be adjusted in real time through an intelligent control system to ensure the best imaging effect.

[0105] The light source consists of an annular LED lamp bead array, a drive circuit, a heat dissipation device, and a diffuser. The LED lamp beads are evenly arranged around the lens. It is mainly used for the detection of objects with planar features, such as the detection of surface components on a circuit board. The light rays are evenly projected onto the object surface from all around, reducing the shadows caused by the height difference of components and highlighting the surface details. Experiments show that after using the annular light source, the imaging clarity of the surface components on the circuit board is increased by 30%, and the defect detection accuracy is improved from 80% to 95%. Its unique annular structure has an excellent imaging effect on the edges and pins of tiny components, can clearly present defects such as poor soldering and short circuits, and effectively reduces the misjudgment and omission rates.

[0106] The stable light source filling reduces the influence of ambient light changes on imaging, enabling the machine vision system to maintain consistent performance in different working environments. Whether in the daytime with sufficient natural light or at night with relatively dim light, the full-field imaging uniform light filling technology can provide stable lighting, ensuring the stability and repeatability of the detection results.

[0107] Under uniform lighting conditions, the camera measures the size and shape of the object more accurately. It avoids measurement errors caused by uneven light, especially in high-precision measurement tasks such as the size detection of automotive parts. Precise light filling can ensure the reliability of the measurement results and meet the strict requirements of industrial production for product quality control.

[0108] The industrial-grade light source filling full-field imaging uniform light filling technology effectively improves the imaging quality and detection accuracy of the industrial vision system through various implementation methods. Different embodiments are applicable to different scenarios, and enterprises can select appropriate solutions according to their own needs. In the future, with the development of technology, the uniform light filling technology will play a greater role in Industry 4.0 and intelligent manufacturing, achieving higher-precision and more intelligent industrial detection and production.

[0109] Finally, it should also be noted that in this text, relationships such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variant are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0110] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed over multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0111] The embodiments of the present application have been described above. The above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other ordinary skill in the art in the technical field to understand the embodiments disclosed herein.

Claims

1. A full-field imaging uniform supplementary lighting method for a conveyor belt, characterized in that, It includes the following steps: Obtain the image of the belt area; Input the belt area image into a pre-trained model to obtain the recognition result output by the pre-trained model. Among them, the pre-trained model is a comprehensive model including a light information extraction and analysis model and an object feature recognition model. The light analysis model is used to identify the position and range of the uneven light area in the belt area image, and the object feature recognition model is used to extract and classify the material and shape features of the objects on the belt; Generate an adjustment plan based on the recognition result and a preset light supplement rule, output it, and complete the light supplement of the belt area based on the adjustment plan.

2. The full-field imaging uniform supplementary lighting method for a conveyor belt according to claim 1, wherein The adjustment plan includes: Dynamically adjust the light source intensity, and change the irradiation angle of the light source array in real time according to the object shape and position; Switch the light supplement mode according to the detection requirements of objects with different materials, and change the optical characteristics through intelligent optical materials; determine the position and angle of the reflector, and construct a reflection system to evenly reflect the light to the target area; And calculate the adjustment value of the light supplement parameter according to the uneven light area, shadow and highlight conditions to achieve uniform light supplement.

3. The full-field imaging uniform supplementary lighting method for a conveyor belt according to claim 1, characterized in that, The processing steps of the light information extraction and analysis model include: Obtain the light information of the area based on the belt area image, and calculate the brightness difference value of each area of the image; Compare the brightness difference value with a preset uniformity threshold to identify the under-lighted area and the over-lighted area; Mark the position and range of the uneven light area, and the uneven light area includes the under-lighted area, the over-lighted area, as well as the shadow and highlight areas.

4. The full-field imaging uniform supplementary lighting method for a conveyor belt according to claim 1, wherein The preset light supplement rules include a light intensity adjustment rule, a light source angle adjustment rule, a light supplement mode switching rule, a light uniform distribution rule, and a real-time image analysis feedback rule.

5. The full-field imaging uniform supplementary lighting method for a conveyor belt according to claim 4, wherein The light intensity adjustment rule and the light source angle adjustment rule are: according to the recognition result of the uneven light area, automatically increase the output power of the corresponding light source in the under-lighted area, and reduce the corresponding light source intensity in the over-lighted area; use the constructed dynamically adjustable light source array, through motor drive and high-precision control circuit, according to the shape and position changes of the target object, change the light irradiation angle and intensity in real time.

6. The full-field imaging uniform supplementary lighting method for a conveyor belt according to claim 4, wherein The light supplement mode switching rule includes: When identifying that the object material is metal and cracks need to be detected, start the light supplement mode integrating high-brightness LEDs and ultraviolet-band gas discharge lamps; When it is recognized that the belt area is in a complex light environment, automatically adjust the light transmittance and color through an electrochromic filter to eliminate the interference of ambient light.

7. The full-field imaging uniform supplementary lighting method for a conveyor belt according to claim 4, characterized in that The light uniform distribution rule includes: When identifying the object as a large workpiece, use multiple mirror groups to reflect the light emitted by the light source multiple times to cover the entire field of view; when identifying the object as a large irregular mechanical part, automatically identify the part contour and adjust the light source array to achieve uniform light supplement.

8. The method for uniformly supplementing light for full-field imaging of a conveyor belt according to claim 4, wherein The real-time image analysis feedback rule includes that when it is recognized that there are defects on the surface of the object package, identify the shadow generated by the wrinkle and adjust the light supplement parameter to eliminate the influence.

9. The full-field imaging uniform supplementary lighting method for a conveyor belt according to claim 1, characterized in that The object feature recognition model is a convolutional neural network model, including: A convolutional layer for performing convolutional operations on the belt area image to extract edge and texture features; A pooling layer for reducing the resolution of the feature map; Residual block, deepening the network depth and solving the gradient vanishing problem to learn high-level features; Fully connected layer, mapping the extracted features to different categories to complete the classification and recognition of the object's material and shape.

10. A full-field imaging uniform supplementary lighting system for a conveyor belt, applicable to the full-field imaging uniform supplementary lighting method for a conveyor belt as shown in any one of claims 1-9, characterized in that, It includes: Image acquisition module, including several image sensors installed in the belt area, used to acquire images of the belt area; Lighting analysis module, used to receive the belt area image and calculate the brightness difference of each area of the image, and output the adjustment instruction of the emission intensity of different light sources in the light source array according to the preset uniformity threshold; Object feature recognition module: used to extract and classify the material and shape features of the object on the belt; Lighting adjustment execution module: used to receive the adjustment instruction of the emission intensity, and through the motor drive and high-precision control circuit, to change the brightness and irradiation angle of each light source in the dynamically adjustable light source array in real time; Supplementary light mode switching module, used to control the combined activation of different types of light sources and adjust the optical characteristics of lenses and filters made of intelligent optical materials; Light reflection control module, used to control the position and angle of each mirror in the reflection system composed of multiple mirrors, and evenly reflect the light emitted by the light source to the target area; Real-time image feedback analysis module, used to identify uneven lighting areas, shadows and highlights, quickly calculate the supplementary light parameters to be adjusted according to the analysis results, and feedback to the lighting adjustment execution module for real-time adjustment; Distributed control module, connecting the light sources, optical components and sensor devices in the supplementary light system into a network, and responsible for coordinating the work of each device based on the main controller.

Images