Object lane occupation identification method and system based on dynamic feature enhancement processing

Abstract

The invention discloses an object lane occupation identification method and system based on dynamic feature enhancement processing, and the method comprises the following steps: S1, real-time environment adaptability image collection: adjusting the exposure and gain setting of a camera according to the illumination intensity and color temperature analysis result of the current environment; s2, dynamic feature enhancement processing: performing real-time feature extraction and enhancement on the environment image captured by the adjusted camera based on a deep learning feature extraction network adjusted by a Gaussian model; and S3, intelligent denoising and accurate identification: based on the extracted and enhanced features of the environment image, performing intelligent denoising processing on the environment image by using the dynamic characteristics of the object. The method has the advantages that by dynamically adjusting exposure, gain setting and image feature extraction parameters of the camera and realizing efficient image de-noising processing, the problems of unstable image quality, inaccurate feature extraction and important detail loss in the de-noising process caused by fixed parameter setting in a dynamic and complex environment in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical fields of image processing and machine vision, and particularly to an object occupation recognition method and system based on dynamic feature enhancement processing. Background Art

[0002] Existing object occupation recognition methods mainly rely on automatic exposure adjustment technology, traditional feature extraction algorithms, and image denoising processing technology. The automatic exposure adjustment technology usually adjusts the camera settings depending on the overall image brightness. However, in rapidly changing lighting conditions, this method may not be able to quickly adapt to environmental changes, resulting in unstable image quality. In addition, feature extraction algorithms often use fixed parameters and are insufficient to cope with changes in environmental dynamics or scene complexity, which limits their application effects in dynamic environments. At the same time, image denoising technology may remove important details in the image while processing noise, especially in low-light conditions or high-dynamic scenes, which will further reduce the accuracy of subsequent image recognition and analysis. Summary of the Invention

[0003] The purpose of the present invention is to provide an object occupation recognition method and system based on dynamic feature enhancement processing, so as to solve the foregoing problems existing in the prior art.

[0004] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0005] An object occupation recognition method based on dynamic feature enhancement processing, comprising the following steps:

[0006] S1. Real-time environment adaptable image acquisition: Adjust the exposure and gain settings of the camera according to the analysis results of the light intensity and color temperature of the current environment;

[0007] S2. Dynamic feature enhancement processing: Based on a deep learning feature extraction network adjusted by a Gaussian model, perform real-time feature extraction and enhancement on the environment image captured by the adjusted camera;

[0008] S3. Intelligent denoising and precise recognition: Based on the features of the environment image after extraction and enhancement, perform intelligent denoising processing on the environment image using the dynamic characteristics of the object.

[0009] Preferably, step S1 is specifically: According to the light intensity and color temperature of the current environment, use the environment adaptable exposure adjustment formula to dynamically adjust the exposure value of the camera; the calculation formula is as follows:

[0010] E = a·L b + c·T d

[0011] Among them, E is the exposure value of the camera; L is the light intensity; T is the color temperature; a and c are the linear adjustment coefficients of the light intensity and the color temperature respectively; b and d are the exponential adjustment coefficients of the light intensity and the color temperature respectively.

[0012] Preferably, step S2 is specifically as follows: The deep learning feature extraction network dynamically adjusts the parameters of the Gaussian function according to the environmental image, and uses the adjusted Gaussian function to extract and enhance the features of the environmental image; the calculation formula is as follows,

[0013]

[0014] Among them, F is the extracted and enhanced feature vector; x is the input environmental image; w1 and w2 are the weights of the two Gaussian functions respectively; u1 and u2 are the means of the two Gaussian functions respectively; s1 and s2 are the standard deviations of the two Gaussian functions respectively.

[0015] Preferably, when the light condition in the input environmental image changes significantly, the deep learning feature extraction network automatically adjusts the weights and means of the Gaussian function to adapt to the new light condition, ensuring that the key information can be accurately captured during the feature extraction process under various environmental conditions.

[0016] Preferably, step S3 is specifically as follows: According to the dynamic characteristics of the objects in the environmental image, the denoising level is dynamically adjusted to effectively remove the noise in the environmental image while retaining the important feature information; the calculation formula is as follows,

[0017] I′ = I - α·∑(N·mask(i))

[0018] Among them, I′ is the denoised environmental image; I is the environmental image; α is the dynamic adjustment coefficient; N is the noise model, which is used to estimate the noise distribution in the environmental image; mask(i) is the mask function automatically generated based on the object characteristics, which is used to distinguish different denoising regions in the environmental image.

[0019] Preferably, intelligent denoising and precise recognition identify the key objects in the environmental image through image analysis, evaluate the dynamic characteristics of these objects, and dynamically change the dynamic adjustment coefficient α according to the dynamic characteristics of the objects, so as to adjust the denoising intensity according to the object state and avoid detail loss caused by excessive denoising.

[0020] Preferably, the mask function is generated according to the position, shape and expected dynamics of the object, ensuring the protection of the important features of the environmental image during the denoising process.

[0021] Preferably, before step S1, it also includes,

[0022] S0, Image Capture and Real-time Environment Analysis: Use a high-dynamic-range camera to capture environmental images in real time, and analyze the light intensity and color temperature of the current environment based on the environmental images.

[0023] Preferably, after step S3, it further includes

[0024] S4, Object Occupying Road Identification: Perform object occupying road identification based on the denoised environmental images.

[0025] The purpose of the present invention also lies in providing an object occupying road identification system based on dynamic feature enhancement processing. The identification system can implement the above-mentioned method. The identification system includes

[0026] Real-time Environment Adaptive Image Acquisition Module: Adjust the exposure and gain settings of the camera according to the analysis results of the light intensity and color temperature of the current environment;

[0027] Dynamic Feature Enhancement Processing Module: Based on a deep learning feature extraction network adjusted by the Gaussian model, perform real-time feature extraction and enhancement on the environmental images captured by the adjusted camera;

[0028] Intelligent Denoising and Precise Identification Module: Based on the extracted and enhanced environmental image features, perform intelligent denoising processing on the environmental images using the dynamic characteristics of the objects.

[0029] The beneficial effects of the present invention are as follows: 1. The REAC technology of the present invention can dynamically adjust the exposure and gain according to the real-time environmental light intensity and color temperature, effectively adapting to rapidly changing lighting conditions. During the outdoor monitoring system test, when using the system of the present invention during the lighting transition from bright to dark, the response time of the image exposure stable adjustment is 20% faster than that of the traditional system, ensuring the continuity and clarity of the images. 2. The DFEP technology of the present invention adapts to different scene changes by dynamically adjusting the parameters of the Gaussian model, which is particularly important when processing images in complex backgrounds or low-light environments. In the experiment, the accuracy rate of the images processed by DFEP in the feature recognition test is about 15% higher than that using the traditional feature extraction method, especially showing excellent performance in identifying small objects and details. 3. The INDR technology of the present invention adjusts the denoising parameters according to the dynamic characteristics of the objects, effectively distinguishing noise and important details, and reducing the detail loss caused by excessive denoising. In the application test of the traffic monitoring system, the comprehensive score of the images processed by the method of the present invention in terms of visual clarity and noise control is 20% higher than that of the traditional denoising technology. 4. The present invention integrates the REAC technology, DFEP technology, and INDR technology, not only improving the performance of a single processing step, but also optimizing the overall process to reduce the possibility of error accumulation and improve the reliability of the application system. The processing delay in high-dynamic scenarios is reduced by about 25% compared with the existing system, significantly improving the processing efficiency. Description of the Drawings

[0030] Figure 1 It is a flowchart of the recognition method in the embodiment of the present invention. Detailed implementation manners

[0031] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific implementation manners described herein are only used to explain the present invention and are not used to limit the present invention.

[0032] In this embodiment, an object occupation recognition method with dynamic feature enhancement processing is provided, which is particularly applicable to the fields of intelligent transportation and autonomous driving. This method significantly improves the adaptability and efficiency of image processing in dynamic and complex environments through the integrated application of real-time environment adaptive image capture (REAC), dynamic feature enhancement processing (DFEP), and intelligent denoising and precise recognition (INDR). The REAC technology automatically adjusts the exposure and gain of the camera to adapt to different lighting and color temperature conditions, ensuring the quality of image data. The DFEP technology adjusts the feature extraction process according to the real-time image quality to optimize the accuracy and recognition efficiency of features. The INDR technology uses an improved denoising algorithm to adjust the denoising level according to the dynamic characteristics of objects, effectively removing noise while retaining key image details. The comprehensive application of the method not only improves the accuracy of image recognition but also enhances the adaptability to complex environments, providing an efficient and reliable image processing solution for intelligent transportation systems and autonomous driving technologies. As Figure 1 shown, the method of the present invention mainly includes the following parts: I. Image capture and real-time environment analysis

[0033] Use a high-dynamic-range camera to capture environmental images in real time, and analyze the light intensity and color temperature of the current environment based on the environmental images, providing basic image data for subsequent real-time environment adaptive image capture.

[0034] II. Real-time environment adaptive image capture

[0035] In applications such as intelligent transportation systems and autonomous driving, the change of environmental light has a great impact on the quality of image capture, especially in the ever-changing outdoor environment. Correct exposure and gain settings are the key to ensuring the validity of image data. Traditional automatic exposure technologies often only adjust based on the overall image brightness and cannot adapt to rapidly changing light conditions, such as the sudden change from shadow to direct sunlight. Therefore, the present invention proposes a real-time environment adaptive image capture (REAC) technology, which can dynamically adjust the exposure and gain of the camera according to the environmental light intensity and color temperature to adapt to complex and changeable lighting conditions, ensuring the continuity and high quality of image capture.

[0036] Real-time Environment Adaptive Image Capture (REAC) automatically adjusts the exposure and gain settings of the camera by monitoring the ambient light intensity and color temperature in real time, ensuring clear images can be captured under different lighting conditions. By dynamically adjusting the exposure value, this technology ensures the high quality of image data, provides high-quality input for subsequent image processing, and reduces the degradation of image quality caused by lighting changes.

[0037] The REAC technology dynamically adjusts the exposure value of the camera by monitoring the ambient light intensity and color temperature in real time using a specific mathematical model. The core of this process is the environment adaptive exposure adjustment formula,

[0038] E = a·L b + c·T d

[0039] where E is the exposure value of the camera, used to adjust the exposure settings of the camera; L is the light intensity, measured in real time by a photosensitive sensor, with the unit of lux (Lux); T is the color temperature, measured by a color temperature sensor, with the unit of Kelvin (K). The color temperature reflects the "warmth" or "coolness" of the light source color and affects the tone and atmosphere of the image; a, b, c, d are preset parameters, and these parameters are adjusted according to the specific performance of the camera and the expected image processing requirements. The selection of these parameters depends on the general characteristics of the light source and the dynamic range of the imaging system.

[0040] a and c are linear adjustment coefficients, which determine the basic influence degree of light intensity and color temperature on the exposure value. b and d are exponential adjustment coefficients, which determine the sensitivity of exposure adjustment to changes in light intensity and color temperature. The adjustment rules for these four parameters are: for example, a larger b value means that a small change in light intensity will cause a large exposure adjustment, which is suitable for use in environments with frequent extreme light changes.

[0041] In practical applications, such as in intelligent transportation monitoring or the imaging systems of autonomous vehicles, the REAC technology can greatly improve the availability of image data at different times (such as from sunrise to sunset) and different weather conditions (such as sunny, cloudy or overcast). By adjusting the exposure in real time, the system can effectively capture sudden situations on the road, such as vehicles quickly entering the field of view and pedestrians suddenly crossing, ensuring that these key information will not be lost due to improper exposure. In addition, appropriate exposure adjustment also helps with subsequent image analysis and processing, such as object recognition and event detection, because image quality directly affects the accuracy and reliability of these advanced functions. Through this technology, not only the adaptability and flexibility of image capture are improved, but also the effects of subsequent image processing and analysis are guaranteed, thus enhancing the practicality and efficiency of the entire system.

[0042] III. Dynamic Feature Enhancement Processing

[0043] Dynamic Feature Enhancement Processing (DFEP) dynamically adjusts the feature extraction process of images through a deep learning model. The DFEP technology can adjust the key parameters of feature extraction according to the real-time image quality and environmental changes. This not only improves the accuracy of feature extraction but also optimizes the response ability to specific scenarios, especially when the lighting conditions and background complexity change.

[0044] Dynamic Feature Enhancement Processing (DFEP) aims to perform real-time feature extraction and enhancement on the input image data through advanced deep learning techniques to adapt to changing environmental conditions such as lighting changes, weather condition changes, and scene dynamics. The key to this technology lies in its ability to dynamically adjust the key parameters in the feature extraction process according to the real-time analyzed image data to optimize the effect of the recognition algorithm, especially to ensure high accuracy and high reliability in object recognition in complex environments.

[0045] The core of the DFEP technology is a deep learning feature extraction network adjusted based on the Gaussian model. This network uses the input image data (x) to extract and enhance features through a dynamically adjusted Gaussian function. The formula is,

[0046]

[0047] where F is the feature vector after extraction and enhancement, which is used for subsequent image analysis and object recognition; x is the input environmental image, which can be a single image pixel value or a set of pixel blocks; w1 and w2 are the weights of the two Gaussian functions respectively, determining the contribution degree of each Gaussian function in the feature vector; u1 and u2 are the means of the two Gaussian functions respectively, determining the central position of feature extraction, usually dynamically adjusted according to the expected position of the target object in the image; s1 and s2 are the standard deviations of the two Gaussian functions respectively, determining the sensitive range of feature extraction. A smaller standard deviation means that only the image data close to the mean will have a greater impact on feature extraction, which is suitable for situations where the features are very obvious locally.

[0048] The parameters (w1, w2, u1, u2, s1, s2) in the DFEP technology are dynamically adjusted based on real-time image analysis. For example, if a significant change in lighting conditions is detected in the input image, the system will automatically adjust the weight and mean parameters to adapt to the new lighting conditions to ensure that the feature extraction process can still accurately capture key information. At the same time, the adjustment of the standard deviation helps the system maintain stable feature recognition performance when the image noise level changes.

[0049] In an intelligent monitoring system, the DFEP technology can greatly improve the monitoring effect in a changing environment (such as urban streets under different weather and lighting conditions). By dynamically adjusting the feature extraction parameters, the system can still effectively identify pedestrians and vehicles on the road under low-light conditions such as rainy days or at night, reducing the cases of misidentification and missed identification. In addition, this technology is also applicable to the environmental perception system of autonomous vehicles, helping the vehicles to better understand the surrounding environment and improving the safety and reliability of autonomous driving. Through this technology, the system can adjust the parameters of the image processing algorithm in real time to adapt to the rapidly changing external environment from sunlight to shadow, from dry to wet, etc., ensuring that the image recognition system can maintain the optimal performance under various conditions.

[0050] IV. Intelligent Denoising and Precise Recognition

[0051] Intelligent Denoising and Precise Recognition (INDR) can effectively remove image noise while retaining key detail information by dynamically adjusting the parameters during the denoising process. Through the real-time analysis of the dynamic characteristics of objects, this technology adjusts the denoising intensity to ensure that key features are not eliminated during the denoising process, thus improving the accuracy and reliability of the recognition algorithm.

[0052] In image processing, denoising is a key step in improving image quality, especially in the fields of intelligent transportation systems and autonomous driving. The effect of denoising directly affects the accuracy of subsequent object recognition and decision-making. Traditional denoising techniques usually use fixed parameters to process all types of noise, which is often not flexible and effective enough when dealing with actual dynamic scenarios. Therefore, the present invention proposes an intelligent denoising and precise recognition technology (INDR), which can dynamically adjust the denoising level according to the dynamic characteristics of objects (such as moving speed and size), so as to more effectively remove noise while retaining important feature information.

[0053] The INDR technology uses the following formula to perform intelligent denoising processing on images:

[0054] I′ = I - α·∑(N·mask(i))

[0055] where, I ′$I_{denoised}$ is the denoised environmental image, which is the processed image data for further analysis and recognition; $I$ is the environmental image, including all visual information directly captured by the camera; $\alpha$ is the dynamic adjustment coefficient, which is adjusted according to the dynamic characteristics of the object (such as moving speed and object size). Its purpose is to adjust the denoising intensity according to the object state to avoid detail loss caused by excessive denoising; $N$ is the noise model, which is a pre-trained model used to estimate the noise distribution in the environmental image; $mask(i)$ is the mask function automatically generated based on object characteristics, used to distinguish different denoising regions in the environmental image. These masks are generated according to the position, shape and expected dynamics of the object to ensure the protection of important features during the denoising process.

[0056] In practical applications, the INDR technology first identifies key objects in the image through real-time image analysis and evaluates the dynamic characteristics of these objects, such as speed and size. For example, for small and fast-moving objects, the $\alpha$ value may be increased during the denoising process to enhance denoising because such objects are easily affected by motion blur. On the contrary, for large and slow-moving objects, the $\alpha$ value may be reduced to retain more details. In addition, the $mask$ function is dynamically generated for each identified object to ensure that the denoising process does not damage the key features of the object.

[0057] In intelligent monitoring systems, such as urban traffic monitoring or vehicle autonomous driving systems, the INDR technology can significantly improve the image quality and recognition accuracy in complex environments (such as at night, in rainy or snowy days). Through intelligent denoising, the system can effectively extract images of key objects such as vehicles and pedestrians from the noisy background, improve the accuracy of the recognition algorithm, and thus better support traffic management and autonomous driving decisions. The implementation of this technology not only improves the visual quality of the image, but also enhances the effect of subsequent image processing steps, ensuring the efficient and safe operation of the entire system.

[0058] V. Object Occupying Road Recognition

[0059] Based on the denoised environmental image, object occupying road recognition is performed.

[0060] In this embodiment, a system for object occupying road recognition based on dynamic feature enhancement processing is also provided. The recognition system can implement the above-mentioned method. The recognition system includes:

[0061] (1) Real-time environmental adaptability image acquisition module: Adjust the exposure and gain settings of the camera according to the analysis results of the current environmental light intensity and color temperature;

[0062] (2) Dynamic feature enhancement processing module: Based on the deep learning feature extraction network adjusted by the Gaussian model, perform real-time feature extraction and enhancement on the environmental image captured by the adjusted camera;

[0063] (3) Intelligent Denoising and Precise Recognition Module: Based on the extracted and enhanced environmental image features, the dynamic characteristics of objects are used to perform intelligent denoising on the environmental images.

[0064] In this embodiment, from REAC to DFEP: The high-quality image data provided by the REAC technology directly supports the feature extraction process of the DFEP technology. The good input image quality makes the feature extraction more accurate, especially under complex lighting conditions. From DFEP to INDR: The feature vectors adjusted by DFEP are more suitable for the INDR technology to perform efficient denoising because the optimized feature extraction ensures the retention of key information and reduces the information loss that may be introduced during the denoising process.

[0065] The combination of the three technologies, REAC, DFEP, and INDR, provides a method for comprehensively optimizing the image processing flow. From image capture to feature extraction, then to denoising and recognition, each step provides a more optimized input for the next step, ultimately achieving a higher image recognition accuracy and processing efficiency than applying each technology alone. In a dynamically changing environment, this integrated method can significantly improve the adaptability and response speed of the system. Especially in autonomous driving and intelligent monitoring systems, this fast and accurate image processing ability is crucial.

[0066] The technical method that integrates real-time environmental adaptability image acquisition, dynamic feature enhancement processing, and intelligent denoising and precise recognition not only supports each other functionally but also interacts to produce new technical effects. This combination method is not a simple stacking of technologies but through the mutual optimization and support of each step, achieving an overall effect that is superior to the sum of individual steps, reflecting prominent substantive features and significant technological progress. This provides an efficient and reliable solution for the field of intelligent image processing, with obvious market application value and broad development prospects.

[0067] By adopting the above technical solutions disclosed in the present invention, the following beneficial effects are obtained:

[0068] The present invention provides an object occupancy recognition method and system based on dynamic feature enhancement processing. The REAC technology of the present invention can dynamically adjust exposure and gain according to the real-time environmental light intensity and color temperature, effectively adapting to rapidly changing lighting conditions. During the outdoor surveillance system test, when using the system of the present invention during the lighting transition from bright to dark, the response time of the image exposure stable adjustment is 20% faster than that of the traditional system, ensuring the continuity and clarity of the image. The DFEP technology of the present invention adapts to different scene changes by dynamically adjusting the parameters of the Gaussian model, which is particularly important when processing images in complex backgrounds or low-light environments. In the experiment, the accuracy rate of the images processed by DFEP in the feature recognition test is increased by about 15% compared with the traditional feature extraction method, especially showing excellent performance in recognizing small objects and details. The INDR technology of the present invention adjusts the denoising parameters according to the dynamic characteristics of the object, effectively distinguishing noise from important details and reducing the detail loss caused by excessive denoising. In the application test of the traffic surveillance system, the comprehensive score of the images processed by the method of the present invention in terms of visual clarity and noise control is increased by 20% compared with the traditional denoising technology. The present invention integrates the REAC technology, DFEP technology and INDR technology, not only improving the performance of a single processing step, but also the optimization of the overall process reduces the possibility of error accumulation and improves the reliability of the application system. The processing delay in high-dynamic scenes is reduced by about 25% compared with the existing system, significantly improving the processing efficiency.

[0069] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. An object occupying road recognition method based on dynamic feature enhancement processing, characterized in that: including the following steps, S1. Real-time environmental adaptability image acquisition: Adjust the exposure and gain settings of the camera according to the analysis results of the current environmental light intensity and color temperature; S2. Dynamic feature enhancement processing: Based on the deep learning feature extraction network adjusted by the Gaussian model, perform real-time feature extraction and enhancement on the environmental images captured by the adjusted camera; S3. Intelligent denoising and accurate recognition: Based on the features of the environmental images after extraction and enhancement, use the dynamic characteristics of objects to perform intelligent denoising processing on the environmental images.

2. The method for identifying object occupancy based on dynamic feature enhancement processing according to claim 1, wherein: Step S1 is specifically as follows. According to the light intensity and color temperature of the current environment, use the environmental adaptability exposure adjustment formula to dynamically adjust the exposure value of the camera. The calculation formula is as follows, E = a·L b + c·T d where E is the exposure value of the camera; L is the light intensity; T is the color temperature; a and c are the linear adjustment coefficients of the light intensity and color temperature respectively; b and d are the exponential adjustment coefficients of the light intensity and color temperature respectively.

3. The object occupancy recognition method based on dynamic feature enhancement processing according to claim 1, wherein: Step S2 is specifically as follows. The deep learning feature extraction network dynamically adjusts the parameters of the Gaussian function according to the environmental image, and uses the adjusted Gaussian function to extract and enhance the features of the environmental image; The calculation formula is as follows, where F is the feature vector after extraction and enhancement; x is the input environmental image; w1 and w2 are the weights of the two Gaussian functions respectively; u1 and u2 are the means of the two Gaussian functions respectively; s1 and s2 are the standard deviations of the two Gaussian functions respectively.

4. The method for identifying object occupation of road based on dynamic feature enhancement processing according to claim 3, wherein: When the light conditions in the input environmental image change significantly, the deep learning feature extraction network automatically adjusts the weights and means of the Gaussian function to adapt to the new light conditions, ensuring that the key information can be accurately captured in the feature extraction process under various environmental conditions.

5. The method for identifying object occupancy based on dynamic feature enhancement processing according to claim 1, wherein: Step S3 is specifically as follows. According to the dynamic characteristics of objects in the environmental image, dynamically adjust the denoising level to effectively remove the noise in the environmental image while retaining important feature information. The calculation formula is as follows, I′ = I - α·∑(N·mask(i)) where I′ is the environmental image after denoising; I is the environmental image; α is the dynamic adjustment coefficient; N is the noise model, which is used to estimate the noise distribution in the environmental image; mask(i) is the mask function automatically generated based on the object characteristics, which is used to distinguish different denoising regions in the environmental image.

6. The method for identifying object occupancy based on dynamic feature enhancement processing according to claim 5, wherein: Intelligent denoising and accurate recognition identify the key objects in the environmental image through image analysis, evaluate the dynamic characteristics of these objects, and dynamically change the dynamic adjustment coefficient α according to the dynamic characteristics of the objects, so as to adjust the denoising intensity according to the object state and avoid detail loss caused by excessive denoising.

7. The method for identifying object occupation of road based on dynamic feature enhancement processing according to claim 5, wherein: The mask function is generated according to the position, shape and expected dynamics of the object to ensure the protection of important features of the environmental image during the denoising process.

8. The object occupancy recognition method based on dynamic feature enhancement processing according to claim 1, characterized in that: Before step S1, it also includes, S0. Image capture and real-time environmental analysis: Use a high-dynamic range camera to capture environmental images in real time, and analyze the light intensity and color temperature of the current environment based on the environmental images.

9. The method for identifying object occupancy based on dynamic feature enhancement processing according to claim 1, wherein: After step S3, it also includes, S4. Object occupancy identification: According to the environmental image after denoising, perform object occupancy identification.

10. An object occupying road recognition system based on dynamic feature enhancement processing, characterized in that: The recognition system can implement the method described in any one of claims 1 to 9 above. The recognition system includes, Real-time environmental adaptability image acquisition module: Adjust the exposure and gain settings of the camera according to the analysis results of the current environmental light intensity and color temperature; Dynamic feature enhancement processing module: Based on a deep learning feature extraction network adjusted by the Gaussian model, perform real-time feature extraction and enhancement on the environmental images captured by the adjusted camera; Intelligent denoising and precise recognition module: Based on the features of the environmental images after extraction and enhancement, perform intelligent denoising on the environmental images using the dynamic characteristics of the objects.

Images