Automatic driving complex scene traffic sign detection method based on multi-channel image extraction

Abstract

The invention discloses an automatic driving complex scene traffic sign detection method based on multi-channel image extraction, and belongs to the technical field of image detection. According to the method, the problem of low detection accuracy caused by image missing, deformation and blurring in an existing automatic driving method is solved, multiple to-be-detected traffic sign images in a complex scene are acquired, enhancement processing is performed on corresponding channel images according to signs to which traffic signs belong, and the detection accuracy is improved. Extracting a feature region from each traffic sign; each to-be-detected traffic sign image is identified and detected through a convolutional neural network to ensure the accuracy of traffic sign detection, and then a plurality of detection results are stored to form a traffic sign database, so that the current to-be-detected traffic sign image is subjected to the above operation, and the traffic sign detection accuracy is improved. Matching with a traffic sign database to obtain a corresponding detection result; and a result is fed back to the automatic driving system, so that the automatic driving system is assisted to control the automobile to run according to the traffic sign.

Description

Technical Field

[0001] The present invention relates to the field of image detection technology, and in particular to a method for detecting traffic signs in complex scenes of autonomous driving based on multi-channel image extraction. Background Art

[0002] With the deep integration of new generation digital technologies such as artificial intelligence and 5G mobile communications with the automotive industry, the intelligent transformation and development of the automotive industry continues to accelerate, and the market's acceptance of autonomous driving continues to increase; traffic sign recognition plays a vital role in intelligent transportation systems, which is directly related to the safe driving of vehicles and the smoothness of road traffic. For unmanned driving systems, the recognition of traffic signs is the basis for their autonomous navigation and decision-making. Unmanned vehicles need to be able to accurately recognize and understand the information of traffic signs in order to make correct driving decisions and ensure driving safety. Improving the accuracy and real-time performance of traffic sign recognition is a key issue that needs to be solved in the process of moving towards practical application. Incorrect recognition results not only cannot play a role in assisting driving, but may also lead to safety accidents. Therefore, the research and development of traffic sign recognition technology is of great significance to improving road traffic safety and promoting the widespread application of intelligent transportation systems.

[0003] In complex traffic scenes, small target detection of traffic signs is a major challenge in the field of target detection. In existing detection technologies, traffic sign images are missing, deformed, and blurred due to factors such as lighting and occlusion, resulting in low accuracy in traffic sign recognition in complex scenes during autonomous driving.

[0004] Therefore, it does not meet the existing needs. For this reason, we propose a traffic sign detection method for complex scenes in autonomous driving based on multi-channel image extraction. Summary of the invention

[0005] The purpose of the present invention is to provide a method for detecting traffic signs in complex scenes in an autonomous driving system based on multi-channel image extraction. The method comprises the following steps: obtaining a plurality of traffic sign images to be detected in a complex scene, enhancing the corresponding channel images according to the signs to which the traffic signs belong, and extracting feature areas from each traffic sign; identifying and detecting each traffic sign image to be detected by a convolutional neural network to ensure the accuracy of traffic sign detection, and then saving a plurality of detection results to form a traffic sign database, so that after the current traffic sign image to be detected is subjected to the above-mentioned operation, it is matched with the traffic sign database to obtain the corresponding detection result; and then feeding back the result to the autonomous driving system to assist the autonomous driving system in controlling the driving of the car according to the traffic signs, thereby solving the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The method for detecting traffic signs in complex scenes of autonomous driving based on multi-channel image extraction includes the following steps:

[0008] Step 1: Acquire images of multiple traffic signs to be detected in a complex scene to form an image sample of the traffic sign;

[0009] Step 2: Separate the multi-channel image of the traffic sign into separate channels, including a red channel, a green channel, and a blue channel, through an image processing library, and assign a corresponding name or identifier to each channel according to the meaning of the channel;

[0010] Step 3: Determine the image sign area of ​​the traffic sign to be detected, enhance the corresponding channel image according to the channel name to which the sign area belongs, and keep the remaining channel images as they are; then extract the channel area image by the color histogram method to obtain the feature area of ​​the traffic sign image to be detected;

[0011] Step 4: Create a convolutional neural network, import the image of the traffic sign to be detected processed in step 4 into the convolutional neural network for training and testing, and detect the features of the traffic sign image to be detected through the convolutional neural network;

[0012] Step 5: perform sign recognition on multiple images of traffic signs to be detected using the above steps, and save the recognition results and corresponding images to form a traffic sign database;

[0013] Step 6. The image of the traffic sign to be detected in the complex scene obtained in real time is processed according to the above steps, and the features of the traffic sign image to be detected are detected by convolutional neural network recognition, and then matched with the traffic sign database, and the corresponding results are output, including: sign type, meaning and corresponding instructions, and the traffic sign detection results are fed back to the autonomous driving system through wireless transmission technology.

[0014] Furthermore, after acquiring the images of multiple traffic signs to be detected in the complex scene in the step 1, the acquired images of the multiple traffic signs to be detected are transmitted to the image processing library via wireless transmission, and the images of the traffic signs to be detected are preprocessed.

[0015] Furthermore, the names assigned to the red channel, the green channel and the blue channel in the step 2 include: the sign graphics, text and edge shapes of the traffic sign in the traffic sign image to be detected.

[0016] Furthermore, in step 3, the corresponding channel image is enhanced according to the channel name to which the feature area belongs, specifically:

[0017] The image of the traffic sign to be detected is read through the image processing library, and the multi-channel histogram equalization method is used to enhance the specified channel image of the traffic sign to be detected in RGB and HIS space to prevent the color distortion of the image of the traffic sign to be detected due to uneven lighting.

[0018] Furthermore, after obtaining the characteristic region of the traffic sign image to be detected in step 3, the following steps are included:

[0019] The traffic sign images to be detected after feature processing are divided into training sets and test sets. The training set is used to establish the feature points of the convolutional neural network, and the test set is used to verify the accuracy of the convolutional neural network.

[0020] Furthermore, after dividing the feature-processed traffic sign images to be detected into a training set and a test set, the following steps are included:

[0021] After extracting features from the multiple traffic sign images to be detected in the above steps, they are saved to form a sample data set; the sample data set includes traffic sign symbols in different environments, different weather conditions, sizes and colors, and different sample data sets are marked and classified according to the content of the traffic signs.

[0022] Furthermore, the step 4 detects the features of the traffic sign image to be detected by using a convolutional neural network, specifically:

[0023] The training set of traffic sign images to be detected is imported into the convolutional neural network for training. The feature area of ​​the training set is identified by the convolution kernel to establish the feature points of the convolutional neural network. The test set is then imported into the convolutional neural network for testing, and the test results are output. The feature area identified by the test results is compared with the feature area extracted in advance. After determining the accuracy of the convolutional neural network test, an optimized convolutional neural network is constructed.

[0024] Furthermore, after the optimized convolutional neural network is constructed, if a new traffic sign image appears, the above steps are followed and the new image is continuously imported into the convolutional neural network for training and testing, and the convolutional neural network is iteratively optimized.

[0025] Furthermore, the image of the traffic sign to be detected is preprocessed, specifically:

[0026] The noise in the image of the traffic sign to be detected is removed, and the image is enhanced, the contrast and brightness of the image are adjusted, and the size of the image is uniformly adjusted to reduce interference factors in the image of the traffic sign to be detected.

[0027] Furthermore, the images of the multiple traffic signs to be detected in the complex scene acquired in the step 1 are all RGB images.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] The present invention obtains images of multiple traffic signs to be detected in a complex scene, separates channels for each image, and then enhances the corresponding channel images according to the sign graphics, text and edge shapes of the traffic signs, so as to facilitate the extraction of feature areas from each traffic sign; further identifies and detects the feature areas of each traffic sign image to be detected through a convolutional neural network, so as to ensure the accuracy of traffic sign detection; and then saves the identification and detection results of multiple traffic sign images to be detected, so as to form a complete traffic sign database; so that after the current traffic sign image to be detected is processed according to the above operation, it is matched with the traffic sign database to obtain the corresponding detection result; and then the result is fed back to the automatic driving system through wireless transmission technology, so as to assist the automatic driving system in controlling the vehicle to drive according to the content of the traffic sign; this method not only improves the safety and reliability of automatic driving vehicles in complex environments, but also helps to improve the intelligence level of automatic driving technology, so that it can better adapt to various road conditions and traffic environments, thereby providing users with a more convenient and safe travel experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the method for detecting traffic signs in complex scenes for autonomous driving based on multi-channel image extraction of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] In order to solve the technical problem that traffic sign images are missing, deformed and blurred due to factors such as lighting and occlusion in existing detection technologies, resulting in low accuracy in traffic sign recognition in complex scenes during autonomous driving, please refer to Figure 1 , this embodiment provides the following technical solutions:

[0033] The method for detecting traffic signs in complex scenes of autonomous driving based on multi-channel image extraction includes the following steps:

[0034] Step 1: Obtain images of multiple existing traffic signs to be detected in complex scenes to form image samples of traffic signs; wherein, the images of multiple traffic signs to be detected in complex scenes are all RGB images; specifically, the method for obtaining images of existing traffic signs can use the existing public traffic sign image dataset, which includes a large number of traffic sign images, such as speed limit, construction, no parking and other signs; or through field photography, photographing traffic signs on actual roads or in simulated environments, this method can obtain more realistic and diverse images, including traffic signs at different angles and lighting conditions. In this way, a richer dataset can be obtained, which helps to improve the generalization ability of the model; or the existing pictures related to traffic signs can be captured from the Internet using web crawler technology, which can be used to train and test machine learning models.

[0035] After acquiring images of multiple traffic signs to be detected in a complex scene, the acquired images of the multiple traffic signs to be detected are transmitted to an image processing library through wireless transmission, and the images of the traffic signs to be detected are preprocessed, specifically:

[0036] Remove noise from the image of the traffic sign to be detected, perform image enhancement, adjust the contrast and brightness of the image, and uniformly adjust the size of the image to reduce interference factors in the image of the traffic sign to be detected; thereby improving image quality and highlighting the features of the traffic sign, helping to reduce the impact of noise, extract the edge information of the image, and prepare for subsequent feature extraction and recognition.

[0037] Step 2: Separate the multi-channel image of the traffic sign into separate channels through the image processing library, including: red channel, green channel and blue channel, and assign a corresponding name or identifier to each channel according to the meaning of the channel; wherein, the names assigned to the red channel, green channel and blue channel include: the sign graphics, text and edge shape of the traffic sign in the traffic sign image to be detected; specifically, in the OpenCV image processing library, the split() function is used to separate a multi-channel image, such as a color image, into separate channels. The split() function accepts a multi-channel image as input and outputs an array containing the separated single-channel image. The following are the basic steps for channel separation using the split() function:

[0038] Use the functions in the cv2 module through the OpenCV library to load the traffic sign image file to be detected, and use the imread() function to read the traffic sign image file to be detected; use the split() function to separate the traffic sign image to be detected into separate channels, so that the multi-channel image can be easily decomposed into separate channels for further processing or analysis.

[0039] In this embodiment, for example, an image of a traffic sign to be detected in a complex scene is obtained through the electronic police system, and after the image is subjected to noise reduction and enhancement processing, the multi-channel image of the traffic sign is separated into separate channels through the split() function in the image processing library, wherein the red channel name represents the sign graphic in the current traffic sign image, the green channel represents the text in the current traffic sign image, and the blue channel is the edge shape of the current traffic sign, thereby achieving regional separation of the content of the current traffic sign image, so that the corresponding channel images can be processed separately according to different requirements, which not only improves the efficiency of image processing, but also highlights the key areas that need to be paid attention to in the image.

[0040] From this description, we can see that after the library uses the split() function algorithm to perform channel separation processing on the traffic sign image, the image shown by the traffic sign is separated into a text channel, a sign shape channel, and a logo graphic channel, so that subsequent enhancement processing can be performed on key channels, thereby improving the accuracy of traffic sign image recognition.

[0041] Step 3: Determine the image sign area of ​​the traffic sign to be detected, enhance the corresponding channel image according to the channel name to which the sign area belongs, and keep the remaining channel images as they are; then extract the channel area image by the color histogram method to obtain the feature area of ​​the traffic sign image to be detected; wherein, enhance the corresponding channel image according to the channel name to which the feature area belongs, specifically:

[0042] The image of the traffic sign to be detected is read through the image processing library, and the multi-channel histogram equalization method is used to enhance the designated channel image of the traffic sign to be detected in RGB and HIS space to prevent the color distortion of the image of the traffic sign to be detected due to uneven lighting;

[0043] Specifically, the RGB space directly corresponds to the three color channels of red, green, and blue, while the HSI space divides the color information into three independent parts: hue, saturation, and brightness. Processing of RGB color space: In the RGB color space, each pixel is defined by the values ​​of the three components of red, green, and blue. Each component can be independently processed with histogram equalization to enhance the contrast and details of the image. The pixel representation of the RGB color space uses 32 bits, of which 8 bits are used for each primary color (red, green, and blue), and the remaining 8 bits can be used as an alpha channel or ignored. Processing of HSI color space: In the HSI color space, histogram equalization can be performed separately for the three channels of hue, saturation, and brightness to better process color and brightness information. Analysis and processing of HSI color space is particularly useful for identifying and separating specific colors, which is particularly important in the color recognition of traffic signs. By performing histogram equalization on the specified channels of the traffic sign image in the RGB and HSI color spaces, the visual effect of the traffic sign image can be effectively enhanced, and the contrast and clarity of the traffic sign image can be improved. This method has a significant effect on improving image quality, especially in low-light or high-color recognition scenes.

[0044] The name of the red channel in the current traffic sign image represents the traffic sign graphic, i.e., the horn and slash area, the green channel represents the traffic sign text, and the blue channel represents the edge shape of the traffic sign, i.e., a circle. According to the main factors of traffic sign detection, the red channel can be used as the key channel image. Based on this, the multi-channel histogram equalization method is used to enhance the red channel image in the current traffic sign image in RGB and HIS space, i.e., image enhancement is performed on the horn and slash area, while the blue channel image and the green channel image remain as they are, thereby highlighting the content of the red channel image and ensuring the clarity of the horn and slash area image.

[0045] After channel separation of the traffic sign image, important channels are identified and enhanced, thereby highlighting the clarity of key areas in the traffic sign image, thereby ensuring the accuracy of subsequent convolutional neural network training, and thereby ensuring the accuracy of the convolutional neural network's automatic recognition and detection of traffic signs in the later stage.

[0046] After obtaining the characteristic region of the traffic sign image to be detected, the following steps are included:

[0047] The image of the traffic sign to be detected after feature processing is divided into a training set and a test set, the training set is used to establish the feature points of the convolutional neural network, and the test set is used to verify the accuracy of the convolutional neural network; after the image of the traffic sign to be detected after feature processing is divided into a training set and a test set, the following steps are included:

[0048] After feature extraction is performed on multiple traffic sign images to be detected according to the above steps, they are saved to form a sample data set; the sample data set includes traffic sign signs in different environments, different weather conditions, sizes and colors, and different sample data sets are marked and classified according to the content of the traffic signs; so as to enrich the feature points of the convolutional neural network and optimize the diversity and accuracy of the convolutional neural network; after marking and classifying different sample data sets according to the content of the traffic signs, corresponding retrieval keywords can also be constructed according to the marked content, so that the required sample data can be quickly retrieved later.

[0049] Step 4: Create a convolutional neural network, import the image of the traffic sign to be detected processed in step 4 into the convolutional neural network for training and testing, and use the convolutional neural network to identify and detect the features of the traffic sign image to be detected, specifically:

[0050] The training set of the traffic sign image to be detected is imported into the convolutional neural network for training, and the feature area belonging to the training set is identified by the convolution kernel to establish the feature points of the convolutional neural network; then the test set is imported into the convolutional neural network for testing, and the test results are output. The feature area identified by the test results is compared with the feature area extracted in advance, and after determining the accuracy of the convolutional neural network test, an optimized convolutional neural network is constructed; after the optimized convolutional neural network is constructed, if a new traffic sign image appears, the above steps are followed, and it is continuously imported into the convolutional neural network for training and testing, and the convolutional neural network is iteratively optimized.

[0051] After processing the traffic sign image to be detected according to the above steps, the red channel is obtained, that is, the horn and slash area is the characteristic area of ​​the sign, and its image is backed up and divided into a training set and a test set. The training set is first imported into the convolutional neural network for training, and the training result is output; then the test set is imported into the trained convolutional neural network for test verification, and the test result is output; then the test result is checked with the training result to confirm whether the two exercise results are consistent, that is, the horn and slash area is the characteristic area; after obtaining the characteristic area, the traffic sign content in the traffic sign image to be detected is obtained according to the text content specified by the traffic management department, that is, no honking.

[0052] Step 5. Use the above steps to perform sign recognition on multiple images of traffic signs to be detected, and save the recognition results and corresponding images to form a traffic sign database; specifically, according to the above embodiment, perform feature detection on multiple traffic sign images in complex scenes acquired by the electronic police system and obtain corresponding text content; then save the multiple results obtained and the corresponding images; further, after constructing the traffic sign database, corresponding search keywords can be established according to the image content or text content of the feature area, so that the same pre-stored content can be retrieved from the traffic sign database according to the feature content of the feature detection result at a later time, so as to quickly identify and detect the content of the traffic sign for automatic driving.

[0053] Step 6: The image of the traffic sign to be detected in the complex scene acquired in real time is processed according to the above steps, and the features of the traffic sign image to be detected are detected by convolutional neural network recognition, and then matched with the traffic sign database, and the corresponding results are output, including: sign type, meaning and corresponding instructions, and the traffic sign detection results are fed back to the autonomous driving system through wireless transmission technology;

[0054] Continuing with the above embodiment, for example, when the automatic driving system is driving a car, it recognizes a traffic sign and first pre-processes it to ensure the clarity of the traffic sign; it performs multi-channel separation on the traffic sign and extracts the sign image of the traffic sign as a feature area, namely, the horn and slash area; after enhancing the channel image, it is imported into a convolutional neural network for recognition and detection, and the detection result, namely, the horn and slash area, is output; then, according to the image content of the area, a search and match is performed in the traffic sign database, such as inputting the words horn / slash to obtain the corresponding pre-stored content, selecting the same sample image as the matching object, and obtaining the text analysis content of the current traffic sign according to the text analysis content corresponding to the matching object, and feeding back the content to the automatic driving system through wireless transmission technology, thereby prompting the automatic driving system to prohibit honking on this road section, thereby realizing the automatic driving complex scene traffic sign detection process based on multi-channel image extraction.

[0055] The beneficial effects achieved by the above content: Through the above operations, not only the safety and reliability of autonomous driving vehicles in complex environments are improved, but also the intelligence level of autonomous driving technology is improved, enabling it to better adapt to various road conditions and traffic environments, thereby providing users with a more convenient and safe travel experience.

[0056] Working principle: The images of multiple traffic signs to be detected in complex scenes are obtained through the electronic police system, the multi-channel images of traffic signs are separated into separate channels through the image processing library, and the corresponding channel images are enhanced according to the characteristic areas, and then the characteristic areas of the traffic sign images to be detected are extracted through the color histogram method, and the features of the traffic sign images to be detected are detected through convolutional neural network recognition; according to the above operations, the images of multiple traffic signs to be detected are recognized and detected in turn, and a traffic sign database is formed; the images of traffic signs to be detected in complex scenes obtained in real time are processed according to the above operations, and the features of the traffic sign images to be detected are detected through convolutional neural network recognition, and matched with the traffic sign database to obtain the corresponding results, and the traffic sign detection results are fed back to the automatic driving system through wireless transmission technology.

[0057] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0058] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A traffic sign detection method for complex scenes in autonomous driving based on multi-channel image extraction, characterized in that: The following steps are involved: Step 1: Acquire images of multiple traffic signs to be detected in a complex scene to form an image sample of the traffic sign; Step 2: Separate the multi-channel image of the traffic sign into separate channels, including a red channel, a green channel, and a blue channel, through an image processing library, and assign a corresponding name or identifier to each channel according to the meaning of the channel; Step 3: Determine the image sign area of ​​the traffic sign to be detected, enhance the corresponding channel image according to the channel name to which the sign area belongs, and keep the remaining channel images as they are; then extract the channel area image by the color histogram method to obtain the feature area of ​​the traffic sign image to be detected; Step 4: Create a convolutional neural network, import the image of the traffic sign to be detected processed in step 4 into the convolutional neural network for training and testing, and detect the features of the traffic sign image to be detected through the convolutional neural network; Step 5: perform sign recognition on multiple images of traffic signs to be detected using the above steps, and save the recognition results and corresponding images to form a traffic sign database; Step 6. The image of the traffic sign to be detected in the complex scene obtained in real time is processed according to the above steps, and the features of the traffic sign image to be detected are detected by convolutional neural network recognition, and then matched with the traffic sign database, and the corresponding results are output, including: sign type, meaning and corresponding instructions, and the traffic sign detection results are fed back to the autonomous driving system through wireless transmission technology.

2. The method for detecting traffic signs in complex scenes for autonomous driving based on multi-channel image extraction according to claim 1 is characterized by: After the images of the multiple traffic signs to be detected in the complex scene are acquired in the step 1, the acquired images of the multiple traffic signs to be detected are transmitted to the image processing library through wireless transmission, and the images of the traffic signs to be detected are preprocessed.

3. The method for detecting traffic signs in complex scenes in autonomous driving based on multi-channel image extraction according to claim 1 is characterized by: The names assigned to the red channel, the green channel and the blue channel in the step 2 include: the sign graphics, texts and edge shapes of the traffic sign in the image of the traffic sign to be detected.

4. The method for detecting traffic signs in complex scenes in autonomous driving based on multi-channel image extraction according to claim 1, characterized in that: In the step 3, the corresponding channel image is enhanced according to the channel name to which the feature area belongs, specifically: The image of the traffic sign to be detected is read through the image processing library, and the multi-channel histogram equalization method is used to enhance the specified channel image of the traffic sign to be detected in RGB and HIS space to prevent the color distortion of the image of the traffic sign to be detected due to uneven lighting.

5. The method for detecting traffic signs in complex scenes in autonomous driving based on multi-channel image extraction according to claim 1 is characterized by: After the characteristic region of the traffic sign image to be detected is obtained in step 3, the following steps are included: The traffic sign images to be detected after feature processing are divided into training sets and test sets. The training set is used to establish the feature points of the convolutional neural network, and the test set is used to verify the accuracy of the convolutional neural network.

6. The method for detecting traffic signs in complex scenes in an autonomous driving environment based on multi-channel image extraction according to claim 5, characterized in that: After the feature-processed traffic sign images to be detected are divided into a training set and a test set, the following steps are included: After extracting features from the multiple traffic sign images to be detected in the above steps, they are saved to form a sample data set; the sample data set includes traffic sign symbols in different environments, different weather conditions, sizes and colors, and different sample data sets are marked and classified according to the content of the traffic signs.

7. The method for detecting traffic signs in complex scenes in an autonomous driving environment based on multi-channel image extraction according to claim 1, characterized in that: The fourth step is to detect the features of the traffic sign image to be detected by using a convolutional neural network, specifically: The training set of traffic sign images to be detected is imported into the convolutional neural network for training. The feature area of ​​the training set is identified by the convolution kernel to establish the feature points of the convolutional neural network. The test set is then imported into the convolutional neural network for testing, and the test results are output. The feature areas identified by the test results are compared with the feature areas extracted in advance. After determining the accuracy of the convolutional neural network test, an optimized convolutional neural network is constructed.

8. The method for detecting traffic signs in complex scenes in an autonomous driving system based on multi-channel image extraction according to claim 7, characterized in that: After the optimized convolutional neural network is constructed, if a new traffic sign image appears, it will continue to be imported into the convolutional neural network for training and testing, and the convolutional neural network will be iteratively optimized.

9. The method for detecting traffic signs in complex scenes in an autonomous driving system based on multi-channel image extraction according to claim 2, characterized in that: The image of the traffic sign to be detected is preprocessed as follows: The noise in the traffic sign image to be detected is removed, and the image is enhanced, the contrast and brightness of the image are adjusted, and the size of the image is uniformly adjusted.

10. The method for detecting traffic signs in complex scenes in autonomous driving based on multi-channel image extraction according to claim 1, characterized in that: The images of the multiple traffic signs to be detected in the complex scene obtained in the step 1 are all RGB images.

Images