Intelligent environment color identification method

Abstract

The invention relates to an intelligent environment color recognition method, which realizes accurate control of a color recognition technology under the working condition of a sea-land dual-purpose sound and temperature hidden transfer device, and comprises the following steps that: an image acquisition module acquires various kinds of environment image data; the preprocessing module optimizes the image quality and the data format; the color feature extraction module extracts multi-dimensional features by using a deep learning algorithm; the deep learning model module learns a mapping relation through mass data training; the color recognition module feeds back and accurately recognizes colors based on the training model and the confidence coefficient; and the result output module displays the result in a plurality of visual modes. The environment color recognition method improves the color recognition accuracy and adapts to complex environment illumination and color combination; the method has good adaptability and is suitable for various scenes; reliable data processing is guaranteed, real-time performance, stability and personalized control are achieved, and user experience and working efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of machine vision technology in sea and land scenes, and in particular to an intelligent environmental color recognition method. Background Art

[0002] In many application scenarios, it is important to accurately identify the color information in the environment. For example, in smart home systems, the lighting effect is automatically adjusted according to the color of the environment; in industrial production, the product color is detected and classified; in artistic creation, the color inspiration of the environment is obtained. However, the existing color recognition technology often has problems such as low accuracy and poor adaptability when facing complex and changeable ambient lighting conditions and diverse color distributions.

[0003] In traditional color recognition technology, methods based on fixed thresholds are mostly used. For example, in some simple image color recognition software, a specific RGB (red, green, blue) color threshold range is set to determine the color category. However, this method has poor accuracy in complex environments, because factors such as changes in environmental light and surface materials of objects can cause visual changes in color, and fixed thresholds are difficult to adapt to these changes. In addition, although some existing intelligent recognition technologies take into account some influencing factors, they lack a comprehensive analysis of the overall characteristics of the environment. For example, some color recognition algorithms based on machine learning may only target a limited set of samples during training, and do not take into account the diversity and complexity of colors in different environmental scenes, resulting in insufficient recognition accuracy in practical applications, especially in environmental scenes with multiple colors mixed or color gradients.

[0004] Moreover, the existing color recognition technology cannot adapt well to environmental changes, which makes it difficult to guarantee the accuracy of the recognition results. In practical applications, misjudgments may occur, such as misclassifying similar colors or failing to correctly recognize colors under complex lighting conditions.

[0005] At the same time, the lack of wide adaptability to different environmental scenarios (such as indoor, outdoor, strong light, weak light, etc.) limits the application of these technologies in various environments and cannot meet diverse needs, such as application in complex industrial environments or natural environments. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides an intelligent environmental color recognition method, which realizes precise control of color recognition technology when the amphibious acoustic and temperature concealed transfer device is in operation, and comprises the following steps:

[0007] Step S1: Image acquisition: controlling the image acquisition device through the computing platform to acquire images in the environment;

[0008] Step S2: Image preprocessing: performing operations such as denoising and enhancement on the collected images to improve image quality;

[0009] Step S3: Color space conversion: convert the image from RGB color space to HSV or Lab color space;

[0010] Step S4: Feature extraction: Calculate the color histogram and color equal moment feature vector on the computing system side to describe the color distribution in the image;

[0011] Step S5: Deep learning model training: Use a large amount of image data with labeled color information to train the convolutional neural network model, and optimize the model parameters on the computing system side of the device;

[0012] Step S6: Color recognition: input the features of the image to be recognized into the trained model to obtain the color category prediction result;

[0013] Step S7: Result output: Output the color recognition result to the surface hidden layer display terminal of the terminal device in a user-friendly form.

[0014] In one embodiment of the present invention, the data collected by the image acquisition device includes but is not limited to color information of different lighting conditions, different object material surfaces, and environmental color data of different seasons and different geographical locations.

[0015] In one embodiment of the present invention, the image preprocessing includes normalizing the collected original color data so that the data is within a uniform value range.

[0016] In one embodiment of the present invention, a convolutional neural network is used for feature extraction, local features of color are extracted through a convolutional layer, and feature compression and dimensionality reduction are performed through a pooling layer to improve the efficiency of feature extraction.

[0017] In one embodiment of the present invention, the color recognition module performs color recognition based on the extracted features using a pre-trained adaptive recognition model. The recognition model is trained based on a large amount of labeled sample data and can perform feedback adjustments based on the confidence level of the recognition results. If the confidence level is lower than a set threshold, the data acquisition module may be required to re-collect data or adjust the parameters of preprocessing and feature extraction.

[0018] In one embodiment of the present invention, the convolutional neural network model in the deep learning model training extracts valuable feature information by performing multi-layer convolution and pooling operations on the input image and data, and classifies or predicts through a fully connected layer. The output result is a category label, probability value or feature vector expressed in digital form.

[0019] In one embodiment of the present invention, the result output is in the form of displaying color names, RGB values, or marking different color areas on the original image.

[0020] The above technical solution of the present invention has the following advantages over the prior art: the environmental color recognition method described in the present invention improves the accuracy of color recognition and can cope with complex environmental lighting and diverse color combinations. Its core feature is that an adaptive algorithm based on deep learning is adopted in the environmental color intelligent recognition system. The algorithm can automatically learn the relationship between environmental factors (such as light intensity, object material reflectivity, etc.) and color features. Through training with a large amount of sample data under different environmental scenes, the algorithm can dynamically adjust the recognition strategy according to the input image or environmental data, thereby improving the accuracy of color recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to make the contents of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings.

[0022] Figure 1 It is a schematic block diagram of the structure of the environment color intelligent recognition system of the present invention;

[0023] Figure 2 It is a schematic diagram of the installation position of the environment color intelligent recognition system and the environment control conversion module in the amphibious sound and temperature concealed transfer device of the present invention;

[0024] Figure 3 It is a flow chart of the connection between the environment color intelligent recognition system and the environment control conversion module of the present invention;

[0025] Figure 4 It is a diagram of the data flow and interaction relationship between the intelligent environment color recognition system of the present invention;

[0026] Figure 5 It is a schematic diagram of data processing and model training of the intelligent environment color recognition method described in the present invention.

[0027] As shown in the figure, 1. Environmental color intelligent recognition system; 2. Environmental control conversion module; 3. Third-layer control module conversion layer; 4. Explosion-proof crystal layer; 5. Computing platform of this equipment; 6. Display end; 7. User terminal; 8. Computing system end; 9. Intelligent recognition and collection of environmental data; 10. Establish training module; 11. System input; 12. Color recognition; 13. Matching algorithm; 14. Data transmission; 15. Instruction execution; 16. Terminal device; 17. Image acquisition device; 18. Image preprocessing; 19. Feature extraction; 20. Deep learning model training. DETAILED DESCRIPTION

[0028] Embodiment 1

[0029] like Figure 5 As shown, this embodiment provides an intelligent environmental color recognition method, which realizes precise control of color recognition technology when the amphibious acoustic and temperature concealed transfer device is in operation, and includes the following steps:

[0030] Image acquisition step: In order to protect the hidden security of the device, the image acquisition device 17 of the device computing platform 5 is responsible for acquiring the image in the environment.

[0031] Specifically, the selection and control of image acquisition equipment: a specially designed image acquisition device is used, which runs under the precise control of the computing platform of the device to ensure that image data in the environment can be stably acquired under various complex environmental conditions. The image acquisition device has high sensitivity and adaptability, and can adapt to different light intensity, ambient temperature, humidity and other conditions, while ensuring the hidden security of the device during the image acquisition process to avoid exposing the device's own position or status due to the acquisition behavior.

[0032] Collection of extensive environmental data: The data collected by image acquisition equipment covers a wide range, including but not limited to the following types of environmental color information. In terms of lighting conditions, it includes the characteristics of the surface color of objects in strong light environments, the changes in color in weak light environments, and the differences in color under light of different color temperatures; for the collection of color information on the surface of object materials, it involves various common and special materials, such as metal, plastic, wood, fabric and other different material surfaces. The color reflection characteristics and subtle changes; and also includes the changes in the color of the natural environment in different seasons, such as the overall environmental color differences brought about by the colorful flowers in spring, the shady trees in summer, the golden leaves in autumn, and the white snow in winter, as well as the environmental color characteristics unique to different geographical locations, such as the earthy yellow tones in desert areas, the blue changes in ocean areas, and the rich green levels in forest areas.

[0033] Image preprocessing step: The image preprocessing 18: performs operations such as denoising and enhancement on the collected images to improve image quality.

[0034] Specifically, image quality optimization operation: a series of comprehensive preprocessing operations are performed on the collected images. First, denoising is performed, and advanced filtering algorithms such as median filtering and Gaussian filtering are used to effectively remove noise points in the image caused by environmental interference, equipment electronic noise and other factors, making the image clearer and smoother. Then, contrast enhancement operation is performed to adjust the grayscale histogram of the image, stretch the grayscale range of the image, and highlight the layering and detail information of the colors in the image, thereby significantly improving the overall quality of the image, making the subsequent color feature extraction more accurate and effective.

[0035] Normalization of original color data: Strict normalization is performed on the collected original color data. Color data collected from different sources and different devices are uniformly mapped to a specific value range, such as normalizing RGB color values ​​to the [0,1] interval or other predetermined standard value ranges. This process ensures that all subsequent color data is analyzed and processed under the same benchmark, eliminating analysis deviations that may be caused by differences in data sources, and greatly improving the accuracy and stability of the entire color recognition process.

[0036] Color space conversion step: The color space conversion: converts the image from the common RGB color space to the HSV or Lab color space that is more suitable for color analysis.

[0037] Specifically, the adaptive selection of color space: according to the needs of color analysis and environmental characteristics, the image is converted from the common RGB color space to the HSV (hue, saturation, brightness) or Lab color space that is more suitable for color analysis. The HSV color space can more intuitively describe the visual attributes of color. Its hue, saturation, and brightness are independent of each other, which is convenient for analyzing the color hue, vividness, and brightness separately. It has obvious advantages especially when dealing with color recognition tasks related to human visual perception. The Lab color space has good uniformity and perceptual consistency, and can more accurately reflect the differences and similarities between colors. It performs well in scenes that require precise comparison of color differences.

[0038] Accurate application of conversion algorithms: Use accurate mathematical conversion algorithms to achieve conversion from RGB to HSV or Lab color space. In the process of converting RGB to HSV, the values ​​of hue, saturation and brightness are calculated according to specific formulas to ensure that the converted color information can accurately represent the color characteristics of the original image in the new color space. For the conversion of RGB to Lab, a complex mathematical model is also used for accurate conversion to retain the key information of color, providing a reliable data basis for subsequent color feature extraction and analysis.

[0039] Feature extraction step: The feature extraction 19: calculates feature vectors such as color histograms and color moments at the computing system end 8 to describe the color distribution in the image.

[0040] Specifically, multi-dimensional feature vector calculation: On the computing system side, advanced computing algorithms are used to calculate multi-dimensional feature vectors such as color histograms and color moments to comprehensively and meticulously describe the color distribution in the image. The color histogram visually displays the distribution ratio of colors in the image by counting the frequency of occurrence of different color values ​​in the image, and can reflect the composition of the main colors in the image. The color moment starts from the statistical characteristics of the image color, calculates the first-order moment (mean), second-order moment (variance) and third-order moment (skewness), etc. of the color, and further describes the concentration, dispersion and symmetry of the color distribution, thereby more deeply describing the color feature information of the image.

[0041] In-depth application of deep learning algorithms: Use deep learning algorithms, especially convolutional neural networks (CNNs), to perform deep feature extraction on preprocessed image data. CNN automatically learns local and global features of color in images by constructing a multi-layer convolution layer and pooling layer structure. The convolution layer uses convolution kernels to slide on the image to perform convolution operations to extract local patterns and texture information of color. Different convolution kernels can learn color features of different scales and directions. The pooling layer performs dimensionality reduction processing on the feature map output by the convolution layer, reducing the amount of data while retaining key feature information, improving the efficiency and robustness of feature extraction, and thus obtaining a more representative and discriminative color feature vector.

[0042] Deep learning model training steps: The deep learning model training 20: uses a large amount of image data with labeled color information to train the convolutional neural network model, and optimizes the model parameters on the computing system end 8 of this device.

[0043] Specifically, large-scale data-driven training: Use massive amounts of image data with carefully labeled color information to train the convolutional neural network model. These image data cover a variety of environmental scenes, object types, lighting conditions, and color changes, ensuring that the model can learn the complex mapping relationship between rich and diverse color features and actual color categories. During the training process, the model continuously adjusts the internal neuron connection weights to minimize the error between the predicted color category and the actual labeled color category, thereby gradually optimizing the model's parameters and improving the model's ability and accuracy in recognizing different colors.

[0044] Model parameter optimization and adjustment: On the computing system side of this device, advanced optimization algorithms, such as stochastic gradient descent (SGD) and its variants Adagrad, Adadelta, Adam, etc., are used to efficiently optimize model parameters. These optimization algorithms dynamically adjust the step size and direction of parameter updates according to the characteristics of the training data and the loss function of the model, accelerate the convergence speed of the model, avoid falling into the local optimal solution, and ensure that the model can achieve a high performance level under large-scale data training and accurately identify color information in various complex environments.

[0045] Color recognition step: The color recognition 12: inputs the features of the image to be recognized into the trained model to obtain the color category prediction result.

[0046] Specifically, feature input and model prediction: the feature vector extracted from the image to be identified through the previous series of processing steps is accurately input into the trained convolutional neural network model. The model deeply analyzes and calculates the input features based on the mapping relationship between the learned color features and color categories to obtain the color category prediction result. The prediction result represents the possibility of different color categories in the form of probability. For example, the model may predict that the probability of a certain area in the image being red is 0.8, the probability of being orange is 0.15, and the probability of being other colors is 0.05, etc. By comparing the probabilities of different color categories, the most likely color category in the image is determined.

[0047] Confidence feedback adjustment mechanism: The color recognition module has the ability to make feedback adjustments based on the confidence of the recognition results. If the confidence of the color category with the highest probability in the prediction result is lower than the preset threshold, such as 0.7 (the threshold can be adjusted according to the specific application scenario and requirements), the color recognition module will start the feedback adjustment mechanism. This mechanism may require the data acquisition module to re-collect image data, or adjust the relevant parameters of the image preprocessing module and the feature extraction module, such as changing the parameters of the denoising algorithm, adjusting the color space conversion method, optimizing the convolution kernel size and number of feature extraction, etc., and then perform color recognition again until the recognition result that meets the confidence requirements is obtained to ensure the accuracy and reliability of color recognition.

[0048] Result output step: The result output is output in a user-friendly form to the user terminal 7 device surface hidden layer display terminal 6 to display the color recognition result.

[0049] Specifically, user-friendly result display: color recognition results are output in a form that is easy for users to understand and accept. This includes clearly displaying color names, such as "red", "blue", "green", etc., on the display end of the hidden layer on the surface of the terminal device, so that ordinary users can intuitively know the main colors in the image; at the same time, the RGB value of the color can be displayed to provide professional users with accurate color numerical information, which is convenient for them to conduct further color analysis and processing; in addition, different color areas can be accurately marked on the original image, and the distribution of colors in the image can be intuitively presented through different marking methods, such as box selection, filling, color boundary outlining, etc., to help users quickly understand the composition and distribution details of the colors in the image, so as to better meet the needs of different users for color recognition results in different application scenarios.

[0050] Embodiment 2

[0051] like Figure 1 and Figure 2 As shown, this embodiment also provides an intelligent environment color recognition system, wherein the installation position of the environment color intelligent recognition system and the environment control conversion module is in the third layer control module conversion layer 3 inside the amphibious sound and temperature concealed transfer device, including the environment color intelligent recognition system 1 and the environment control conversion module 2; the environment color intelligent recognition system is embedded in the third layer control module conversion layer 3 and is connected;

[0052] like Figure 3 As shown, the connection process between the environment color intelligent recognition system and the environment control conversion module includes:

[0053] First, the amphibious acoustic and temperature concealed transfer device is started, and then the third-layer control module conversion layer 3 of the equipment simultaneously starts the environment color intelligent recognition system 1 and the environment control conversion module 2 in the third-layer control module conversion layer, and connects the environment control conversion module 2 with the explosion-proof crystal layer 4;

[0054] Start the explosion-proof crystal layer 4 and connect it to the environment color intelligent recognition system 1 through the third layer control module conversion layer 3; intelligently recognize and collect the environment data 9 and analyze it;

[0055] Execute the acquisition and establish the deep learning of the training module 10 for the environment, capture the difference of the environment color and compare it with the environment color difference verification module,

[0056] System input 11, the environment color intelligent recognition system continuously obtains image data of the environment as input.

[0057] Color recognition and analysis 12. The environmental color intelligent recognition system uses image processing technology and algorithms to extract and analyze color features of input images.

[0058] Through machine learning models or pattern matching algorithms13, the main color distribution, hue, saturation and other information in the environment are determined.

[0059] Data transmission 14, the recognition system encapsulates the analyzed color data in a standardized data format.

[0060] The packaged data is transmitted to the environment control conversion module 2 via a high-speed data interface (such as Ethernet, USB, etc.).

[0061] The conversion module processes and the environment control conversion module receives data from the color intelligent recognition system.

[0062] According to preset rules and mapping relationships, the color data is converted into corresponding control instructions.

[0063] Control instruction generation, based on the correspondence between color data and control strategy, generates specific environmental control instructions, such as instructions for adjusting the surface color and brightness of the explosion-proof crystal layer 4;

[0064] Verify the validity and security of the generated control instructions;

[0065] Instruction execution 15, the control instruction that passes the verification is sent to the corresponding user-side environmental control explosion-proof crystal layer surface device;

[0066] The environmental control device receives instructions and performs corresponding actions to adjust the environment.

[0067] Feedback and optimization,After the environmental control device performs an action, it feeds back the execution result to the environmental control conversion module.

[0068] The conversion module evaluates the control effect based on the feedback information and optimizes and adjusts the control strategies and rules if necessary to improve the performance and accuracy of the system.

[0069] The feedback information data is output to the terminal device 16.

[0070] Through the above connection process, the environmental color intelligent recognition system can work together with the environmental control conversion module to achieve intelligent perception and analysis of the environmental color and to accurately and effectively control the environment.

[0071] Furthermore, if Figure 4 As shown in the figure, the technical solution of the intelligent environment color recognition system architecture includes:

[0072] 1. Environmental color intelligent recognition system and through the computing platform of this device, the deep learning model module in the computer vision and intelligent recognition system adopts convolutional neural network (CNN) to show powerful image recognition and data analysis capabilities; among them,

[0073] 2. The environmental color intelligent recognition system includes an environmental control conversion module to achieve effective connection and conversion between the two to build an intelligent environmental control system;

[0074] 3. The environmental control conversion module is responsible for converting these analysis results into actual environmental control actions.

[0075] 4. The data acquisition module 17 (image acquisition device): is used to acquire image data in different environmental scenes or directly acquire color data in the environment. The acquired data includes but is not limited to color information of different lighting conditions (strong light, weak light, different color temperatures, etc.), different object material surfaces, and environmental color data in different seasons and different geographical locations.

[0076] 5. The preprocessing module 18 (image preprocessing): preprocesses the collected data, including operations such as image denoising and contrast enhancement, and normalizes the collected original color data to make it within a uniform numerical range for subsequent analysis and processing.

[0077] 6. The feature extraction module 19 (color feature extraction): uses a deep learning algorithm to extract multi-dimensional color features from the preprocessed data. This module uses advanced neural network structures such as convolutional neural networks (CNNs), which can automatically learn the deep-level features of color and features related to the environment, such as extracting local features of color through convolutional layers, and performing feature compression and dimensionality reduction through pooling layers to improve the efficiency of feature extraction. Convolutional neural networks extract valuable feature information by performing multi-layer convolution and pooling operations on input images and data, and classify or predict through fully connected layers. These output results are usually category labels, probability values, or feature vectors expressed in digital form.

[0078] In order to connect the output of CNN with the environmental control conversion module, it is necessary to first define clear conversion rules and mapping relationships. For example, if the output of CNN is the classification results of different environmental states (such as "bright", "dim", and "moderate"), then in the conversion module, these classifications can be mapped to specific environmental control actions (such as "bright" corresponds to dimming the lights, "dim" corresponds to brightening the lights, and "moderate" maintains the current state).

[0079] 7. The recognition module: performs color recognition using a pre-trained adaptive recognition model based on the extracted features. This recognition model is trained based on a large amount of labeled sample data and can accurately determine the color category based on the input features. At the same time, the recognition module can also make feedback adjustments based on the confidence level of the recognition results. If the confidence level is lower than the set threshold, the data acquisition module can be required to re-collect data or adjust the parameters of preprocessing and feature extraction.

[0080] 8. The feedback optimization module: Based on the comparison between the actual recognition results and the real results, the entire recognition system is optimized. If there is a recognition error or the accuracy rate drops, this module will adjust the parameters of the deep learning algorithm and retrain the recognition model to continuously adapt to the new environment and data changes.

[0081] The environmental color intelligent hiding system described in this embodiment improves the accuracy of color recognition and can cope with complex environmental lighting and diverse color combinations. Its core feature is that an adaptive algorithm based on deep learning is adopted in the environmental color intelligent recognition system. The algorithm can automatically learn the relationship between environmental factors (such as light intensity, object material reflectivity, etc.) and color features. Through training with a large amount of sample data in different environmental scenes, the algorithm can dynamically adjust the recognition strategy according to the input image or environmental data, thereby improving the accuracy of color recognition.

[0082] It has good adaptability and can be applied to different scenarios and fields. Its multi-dimensional analysis improves the accuracy and adaptability of intelligent recognition, and builds a multi-dimensional environmental color feature model in its environmental color intelligent recognition system. This model not only takes into account the basic properties of color (such as RGB value, hue, saturation, brightness, etc.), but also integrates the spatial characteristics of the environment (such as the position of the object, the color distribution of surrounding objects, etc.) and time characteristics (such as the changing rules of light in different time periods, etc.). Through this multi-dimensional comprehensive analysis, the environmental color can be identified more accurately, especially in complex color mixing and gradient scenes.

[0083] To ensure data accuracy, in terms of data transmission, the output data of CNN needs to be passed to the environmental control conversion module in a standardized and parseable format. This can be achieved through common data interfaces and communication protocols to ensure data accuracy and timeliness.

[0084] To ensure real-time performance and response speed, environmental control often requires timely response. Therefore, during the connection process, it is necessary to optimize the data processing and conversion process and reduce delays to ensure that environmental control actions can be quickly adjusted according to the analysis results of CNN.

[0085] To improve the stability and reliability of the connection, error handling and abnormal response mechanisms should also be added to the system. For example, if the output of the CNN is abnormal or unrecognizable, the conversion module should have a default processing method to avoid confusion in environmental control.

[0086] It helps improve user experience and work efficiency, and its intelligent applications provide reliable color information support.

[0087] In short, the connection and conversion of the deep learning model module using convolutional neural network and the environmental control conversion module in the environmental color intelligent recognition system is a comprehensive process involving data format standardization, conversion rule definition, real-time optimization and exception handling. Through careful design and effective implementation, the intelligent analysis ability of CNN can be fully utilized to achieve accurate and efficient environmental control.

[0088] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims

1. An intelligent environmental color recognition method, which realizes precise control of color recognition technology when a marine and amphibious acoustic and temperature concealed transfer device is in operation, and is characterized in that: The steps include: Step S1: Image acquisition: controlling the image acquisition device through the computing platform to acquire images in the environment; Step S2: Image preprocessing: performing denoising and enhancement operations on the collected images to improve image quality; Step S3: Color space conversion: convert the image from RGB color space to HSV or Lab color space; Step S4: Feature extraction: Calculate the color histogram and color moment feature vector on the computing system side to describe the color distribution in the image; Step S5: Deep learning model training: Use a large amount of image data with labeled color information to train the convolutional neural network model, and optimize the model parameters on the computing system side of the device; Step S6: Color recognition: input the features of the image to be recognized into the trained model to obtain the color category prediction result; Step S7: Result output: Output the color recognition result to the surface hidden layer display terminal of the terminal device in a user-friendly form.

2. The environmental color recognition method according to claim 1, characterized in that: The data collected by the image acquisition device includes, but is not limited to, color information of surfaces of different object materials under different lighting conditions, and environmental color data of different seasons and different geographical locations.

3. The environmental color recognition method according to claim 1, characterized in that: Image preprocessing includes normalizing the collected raw color data to make it within a uniform value range.

4. The environmental color recognition method according to claim 1, characterized in that: The feature extraction adopts convolutional neural network, which extracts local features of color through the convolution layer, and performs feature compression and dimensionality reduction through the pooling layer to improve the efficiency of feature extraction.

5. The environmental color recognition method according to claim 1, characterized in that: The color recognition module performs color recognition based on the extracted features using a pre-trained adaptive recognition model. The recognition model is trained based on a large amount of labeled sample data and can make feedback adjustments based on the confidence level of the recognition results. If the confidence level is lower than the set threshold, the data acquisition module is required to re-collect data or adjust the parameters of preprocessing and feature extraction.

6. The environmental color recognition method according to claim 1, characterized in that: The convolutional neural network model in deep learning model training extracts valuable feature information by performing multi-layer convolution and pooling operations on the input image and data, and classifies or predicts through the fully connected layer. The output result is a category label, probability value or feature vector expressed in digital form.

7. The environmental color recognition method according to claim 1, characterized in that: The output result may be in the form of displaying color names, RGB values, or marking different color areas on the original image.

Images