Signal lamp recognition method and device, vehicle and medium

Abstract

The application provides a signal lamp identification method, device and equipment and a medium. The method obtains a signal lamp image in a first format and a signal lamp image in a second format collected by an image collection device. The signal lamp image in the second format has M channels. Then, the signal lamp image in the first format is parsed into a signal lamp image with N channels. The signal lamp image in the second format with M channels and the signal lamp image with N channels are combined into a to-be-identified signal lamp image with P channels. Finally, the signal lamp in the to-be-identified signal lamp image is identified to obtain a signal lamp identification result. In the to-be-identified signal lamp image with P channels, the channel colors of the P channels can reflect sufficient light information, so that the signal lamp can be accurately identified in the signal lamp image in a night scene, and the identification accuracy is improved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a method, device, vehicle, and medium for recognizing traffic lights. Background Technology

[0002] Autonomous driving requires the ability to recognize traffic lights on the road.

[0003] In related technologies, the acquisition device can output the acquired traffic light images and identify the traffic lights in the images based on a neural network model.

[0004] However, the RGB format traffic light images output by the acquisition device are affected by light in some driving scenarios, such as nighttime scenarios, resulting in poor image quality of the traffic lights, which may be unrecognizable to the human eye. Summary of the Invention

[0005] To address or partially address the problems existing in related technologies, this application provides a method, apparatus, vehicle, and medium for recognizing traffic lights, which can improve the recognition accuracy of traffic lights in traffic light images.

[0006] The first aspect of this application provides a method for identifying traffic lights, comprising:

[0007] Obtain a traffic light image in a first format and a traffic light image in a second format acquired by an image acquisition device; wherein the traffic light image in the second format has M channels;

[0008] The traffic light image in the first format is parsed into a traffic light image with N channels; where M and N are both positive integers.

[0009] The second-format traffic light image with M channels and the traffic light image with N channels are merged into a traffic light image to be identified with P channels;

[0010] The traffic lights in the image of the traffic lights to be identified are identified to obtain the traffic light identification result.

[0011] A second aspect of this application provides a traffic light identification device, comprising:

[0012] The acquisition module is used to acquire a first-format traffic light image and a second-format traffic light image acquired by the image acquisition device; wherein the second-format traffic light image has M channels;

[0013] The parsing module is used to parse the traffic light image in the first format into a traffic light image with N channels; wherein M and N are both positive integers;

[0014] The merging module is used to merge the second-format traffic light image with M channels and the traffic light image with N channels into a traffic light image to be identified with P channels;

[0015] The recognition module is used to identify the traffic lights in the image of the traffic lights to be identified and obtain the traffic light recognition result.

[0016] A third aspect of this application provides a car, comprising:

[0017] Processor; and

[0018] The memory stores executable code, which, when executed by the processor, causes the processor to perform the methods described above.

[0019] A fourth aspect of this application provides a computer-readable storage medium having executable code stored thereon, which, when executed by a processor of an electronic device, causes the processor to perform the method described above.

[0020] This application provides a method, apparatus, device, and medium for identifying traffic lights. The method obtains a traffic light image in a first format and a traffic light image in a second format acquired by an image acquisition device. The second format traffic light image has M channels. Then, the first format traffic light image is parsed into a traffic light image with N channels. The second format traffic light image with M channels and the N channel traffic light image are merged into a traffic light image to be identified with P channels. Finally, the traffic lights in the traffic light image to be identified are identified to obtain the traffic light identification result. Because the channel information of the P channels in the traffic light image to be identified can reflect sufficient light information, the identification method is robust and can accurately identify traffic lights in low-light scenes such as nighttime scenes, thus improving the identification accuracy.

[0021] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0022] The above and other objects, features and advantages of this application will become more apparent from the description of exemplary embodiments of this application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of this application.

[0023] Figure 1 This is a flowchart illustrating a traffic light identification method according to an embodiment of this application.

[0024] Figure 2This is a flowchart illustrating a traffic light identification method in a real-world scenario, as shown in an embodiment of this application.

[0025] Figure 3 This is a schematic diagram of the arrangement of photosensitive elements shown in the embodiments of this application.

[0026] Figure 4 This is a schematic diagram illustrating the process of synthesizing a P-channel image to be recognized in a scenario shown in an embodiment of this application.

[0027] Figure 5 This is a schematic diagram of the structure of a traffic light identification device shown in an embodiment of this application.

[0028] Figure 6 This is a schematic diagram of the structure of a car shown in an embodiment of this application. Detailed Implementation

[0029] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to make this application more thorough and complete, and to fully convey the scope of this application to those skilled in the art.

[0030] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0031] It should be understood that although the terms "first," "second," "third," etc., may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0032] This application embodiment can be applied to autonomous driving scenarios, especially those requiring traffic light recognition. The traffic lights in this embodiment can be traffic lights, i.e., lights that indicate traffic flow, typically displaying red, green, and yellow colors, and may also include directional information and countdown timers. Image acquisition devices typically output RGB format images with three channels. These three channels are significantly affected by ambient lighting and image signal processing algorithms, making it difficult to accurately identify traffic lights in poorly imaged traffic light images. Therefore, this application embodiment provides a traffic light recognition method, apparatus, device, and medium to improve the accuracy of traffic light recognition in special scenarios.

[0033] The embodiments of this application will be described in detail below.

[0034] Please see Figure 1 , Figure 1 This is a flowchart illustrating a traffic light identification method according to an embodiment of this application.

[0035] This application provides a method for identifying traffic lights, which includes the following steps.

[0036] S101. Obtain a first-format traffic light image and a second-format traffic light image acquired by the image acquisition device; wherein, the second-format traffic light image has M channels.

[0037] In this embodiment, the mobile device may be equipped with an image acquisition device capable of capturing traffic light images in real time. The mobile device may include vehicles, robots, aircraft, etc. The image acquisition device may be, for example, a vehicle-mounted camera or vehicle-mounted radar capable of capturing images in real time. The traffic light image may include road images or scene images showing the environment in which the traffic light is located.

[0038] The image acquisition device acquires a traffic light image in a first format. This first format traffic light image may include light source signals captured by the image sensor of the image acquisition device. It may include more image information, such as physical information about the light intensity and color of the scene, more color depth information, and unprocessed photosensitivity information. In application, the first format may be, for example, RAW format.

[0039] After the image acquisition device acquires a traffic light image in the first format, it can convert the image to a second format as the output. The second format traffic light image can be, for example, a three-channel image, and the format itself can be RGB. The second format traffic light image has broad compatibility and real-time visualization, requires less data than the first format image, and is easier to process. It is understood that in this embodiment, "channel" can refer to a color channel, with each channel representing a color component.

[0040] In applications, image acquisition code can be used to obtain a first-format traffic light image captured by an image acquisition device without format conversion, and a second-format traffic light image after format conversion. The second-format traffic light image can also be acquired simultaneously when the image acquisition device outputs a second-format traffic light image to other models; that is, the second-format traffic light image can be used for parallel output to multiple autonomous driving algorithm models, recognition models, and other algorithms or models.

[0041] It is understandable that the second format traffic light image can also be obtained by converting the first format traffic light image after it has been acquired by the device that acquired the first format traffic light image.

[0042] In this embodiment of the application, obtaining a first-format traffic light image and a second-format traffic light image captured by an image acquisition device may include the following steps.

[0043] S1011. Obtain the first format signal light image acquired by the image acquisition device based on the image acquisition code.

[0044] S1012. Obtain a second format traffic light image output by the image acquisition device after converting the first format traffic light image.

[0045] In this embodiment, since the unprocessed first-format traffic light image acquired by the image acquisition device is typically not directly output as the acquisition result, an image acquisition code is provided. This code can acquire the first-format traffic light image from the image acquisition device. When the image acquisition device converts the first-format traffic light image to a second-format image as the output result, this output result, i.e., the second-format traffic light image, can be directly obtained. For example, the second-format traffic light image can be obtained based on the input address or upload address of the image acquisition device.

[0046] It is understandable that different types of image acquisition devices with different chips can have different image acquisition codes pre-set. The system can automatically match the code based on the chip type.

[0047] S102. Parse the traffic light image in the first format into a traffic light image with N channels; where M is a positive integer greater than two and N is a positive integer greater than three.

[0048] In this embodiment, after obtaining a traffic light image in a first format and a traffic light image in a second format, the traffic light image in the first format is parsed to obtain a traffic light image with N channels corresponding to the first format traffic light image. Here, N has more than three channels, and M has more than two channels.

[0049] In this embodiment, the first format signal light image that has not undergone format conversion contains the original photosensitive data and shooting metadata of the image acquisition device at the time of shooting, such as white balance, ISO settings, etc., and has a high color depth and dynamic brightness range.

[0050] In applications, for example, a traffic light image in the first format has a size of H*W*1, where H is the height, W is the width, and 1 is the number of channels. After parsing, each Bayer unit is grouped into a 1*4 vector, and the image size becomes (H / 2)*(W / 2)*4. Here, the Bayer units can be arranged in a 2x2 matrix in the image. Each Bayer unit contains one red filter, one blue filter, and two green filters, thus enabling the conversion of a single-channel traffic light image in the first format into a traffic light image with four channels.

[0051] In this embodiment of the application, the traffic light image in the second format, such as the traffic light image in RGB format, has N channels after format conversion. For example, the traffic light image in RGB format has three channels.

[0052] S103. The second format traffic light image with M channels and the traffic light image with N channels are merged into a traffic light image to be identified with P channels.

[0053] In this embodiment of the application, in order to improve the recognition accuracy of traffic lights in special scenarios, N channels and M channels are merged into M+N channels to obtain P channels of traffic light images to be recognized.

[0054] It should be noted that the merging in this embodiment does not mean superimposing the same channels. Each of the P channels represents a color space. For example, in this embodiment, the red channel in the N-channel traffic light image and the red channel in the second format traffic light image are respectively used as two channels in the P channels, without superimposing them.

[0055] S104. Identify the traffic lights in the image of the traffic lights to be identified, and obtain the traffic light identification result.

[0056] In this embodiment, after obtaining the image of the traffic light to be identified, which has P channels, the traffic lights in the image with P channels are identified to obtain the traffic light identification result. The traffic light identification result may include traffic light-related information such as the position, color, direction, and / or countdown of the traffic light. The process of identifying the signal can be implemented in various ways.

[0057] In this embodiment, since the channel colors of the P channels in the traffic light image to be identified can reflect sufficient light information, the traffic lights can be accurately identified in the traffic light image even in nighttime scenes, thus improving the recognition accuracy.

[0058] The foregoing embodiments described the process of identifying traffic lights in an image of a traffic light to be identified. This process will be described in detail below.

[0059] In this embodiment, since the traffic light image with P channels has M+N channels, a recognition neural network model can be preset. This recognition neural network model can be a convolutional deep neural network model, which can use M+N channels as input parameters. This convolutional neural network model can be obtained by training the model based on the traffic light image with P channels.

[0060] Understandably, when a convolutional neural network model uses images in the first format during training, it directly uses those images for training. If the image format is second, it is converted to a pseudo-first format, and then used for training. The pseudo-first format can have the same number of channels as the first format, reproducing a certain amount of information from the original scene, and has a wider range of lighting conditions compared to the second format. In applications, an inverse ISP (Imaging Signal Processing) algorithm can be used to convert the second-format image to the pseudo-first format.

[0061] In applications, the process of converting an image to pseudo-first format can be to first downsample the second format image, then add noise and use the inverse ISP algorithm to convert it to a pseudo-first format image.

[0062] It is understandable that the image in the second format cannot be completely restored to the image in the first format. Therefore, in this embodiment, a pseudo-first format image is used to replace the first format image for the data processing process in this embodiment, which also has a good effect.

[0063] In this embodiment of the application, in addition to recognizing the signal light recognition result through a preset recognition neural network model, a recognition algorithm can also be used for recognition.

[0064] In applications, object detection algorithms can be used to identify traffic lights in images. First, the traffic light features, such as color, texture, and shape, can be extracted from the image. Then, these features are matched against preset traffic light features to obtain the matching result. Alternatively, the image can be divided into color regions, and the traffic light frame can be identified based on these color regions, yielding the final traffic light identification result.

[0065] In applications, the recognition neural network model may be one of multiple functional models in autonomous driving. These multiple functional models may also use the second-format traffic light image. If the second-format traffic light image is only used as input to the recognition neural network model, while other functional models still need to obtain the second-format traffic light image again as input, it will lead to a decrease in the overall inference speed and an increase in memory usage for autonomous driving. Therefore, in this embodiment, the second-format traffic light image is used as parallel input to other functional models, achieving reuse and saving system resources for the entire autonomous driving system.

[0066] In this embodiment, when an image acquisition instruction is received from another target functional model with different functions than the recognition neural network model, the second-format traffic light image is used as input to the target functional model and sent to it. This allows the target functional model to reuse the input of the traffic light recognition in this embodiment, eliminating the need for the target functional model to re-acquire the second-format traffic light image. Thus, the target functional model can achieve its functions based on the second-format traffic light image. For example, functions such as moving object recognition and drivable area recognition can be implemented, enabling the target functional model to input its functional results.

[0067] As can be seen, the embodiments of this application are capable of processing P channels of traffic light images to be identified and M channels of traffic light images in a second format in parallel.

[0068] Please see Figure 2 , Figure 2 This is a flowchart illustrating a traffic light identification method in a real-world scenario, as shown in an embodiment of this application.

[0069] Figure 2In step 201, the sensor detects light. In step 202, the sensor generates a traffic light image in a first format, such as a raw image. Then, in step 203, it is parsed into a traffic light image with four channels. In step 204, it is converted into a traffic light image in a second format, such as an RGB image, with three channels using the ISP algorithm. Then, in step 205, it is merged into seven channels as input to a deep convolutional neural network. In step 206, it is input into the deep neural network model, and the traffic light recognition result of the convolutional neural network is obtained, including the traffic light position, color, direction, and countdown.

[0070] In this embodiment, the arrangement of photosensitive units in the first format traffic light image is achieved through Bayer units. The first format traffic light image is composed of multiple Bayer units, each Bayer unit corresponding to N color channels. Through the arrangement of photosensitive elements, N channels can be separated, and then a traffic light image with N channels can be generated based on the N channels.

[0071] In this embodiment, the P-channel traffic light images to be identified can be synthesized from N channels and M channels based on an image array.

[0072] In this embodiment, a blank image array can be created first, and then the array data corresponding to the M channels of the traffic light image with M channels can be written into the first M spaces of the image array. Then, the array data corresponding to the N channels of the traffic light image with N channels can be written into the N spaces after the M spaces to obtain an image array with P channels. Finally, an image of a traffic light to be identified with P channels can be generated based on the image array with P channels.

[0073] Please see Figure 3 , Figure 3 This is a schematic diagram of the photosensitive element arrangement shown in an embodiment of this application. The photosensitive elements are arranged in a fixed manner to acquire traffic light images. The diagram shows four bayer units, each arranged based on the RCCB color channel. In any bayer unit, 31 is the red channel, 32 is the first transparent channel, 33 is the second transparent channel, and 34 is the blue channel. In applications, for example, a RAW format traffic light image is parsed into these four RCCB channels (red, transparent, transparent, and blue), and then merged with the RGB format traffic light image's RGB channels (red, green, and blue), resulting in seven channels: red, transparent, transparent, blue, red, green, and blue. Because the RCCB channels in the RAW format contain rich light information, combined with the brightness and detail of the G channel in the RGB format, even in nighttime scenes where the traffic light produces an aperture, a relatively accurate traffic light recognition result can be obtained. Please continue reading... Figure 4 , Figure 4 This is a schematic flowchart illustrating the process of synthesizing a P-channel image to be identified in a scenario according to an embodiment of this application. Wherein, 401 represents a traffic light image in a first format, 402 represents a traffic light image with N channels obtained by parsing the first format traffic light image, 403 represents a traffic light image in a second format with M channels converted from the first format traffic light image, and 404 represents the synthesis of a traffic light image with P channels to be identified.

[0074] Corresponding to the aforementioned application function implementation method embodiments, this application also provides embodiments of a traffic light identification method, device, vehicle, and medium.

[0075] Figure 5 This is a schematic diagram of the structure of a traffic light identification device shown in an embodiment of this application.

[0076] See Figure 5 This application provides a traffic light identification device 50, comprising:

[0077] The acquisition module 51 is used to acquire a first-format traffic light image and a second-format traffic light image acquired by the image acquisition device; wherein the second-format traffic light image has M channels;

[0078] The parsing module 52 is used to parse the traffic light image in the first format into a traffic light image with N channels;

[0079] The merging module 53 is used to merge the second-format traffic light image with M channels and the traffic light image with N channels into a traffic light image to be identified with P channels.

[0080] The recognition module 54 is used to identify the traffic lights in the traffic light image to be identified and obtain the traffic light recognition result.

[0081] Module 51 is obtained for:

[0082] The first-format traffic light image acquired by the image acquisition device is obtained based on the image acquisition code;

[0083] The image acquisition device outputs a second format signal light image after converting the first format signal light image.

[0084] Recognition module 54 is used for:

[0085] The image of the traffic light to be identified is input into the recognition neural network model to obtain the traffic light recognition result of the traffic light recognition neural network model in the image of the traffic light to be identified;

[0086] The traffic light recognition neural network model is obtained by training a model based on traffic light images with P channels.

[0087] The device further includes a parallel forwarding module for:

[0088] Receive an image acquisition command for a target functional model that has a different function from the recognition neural network model;

[0089] The traffic light image in the second format is sent to the target function model so that the target recognition model can perform the function corresponding to the target function model based on the traffic light image in the second format.

[0090] In some embodiments, parsing the traffic light image in the first format into a traffic light image with N channels includes:

[0091] The first format of the traffic light image is analyzed to obtain the arrangement of the photosensitive elements;

[0092] Based on the arrangement of the photosensitive elements, the single channel of the first format signal light image is separated into N channels;

[0093] Generate a signal light image with N channels based on the N channels.

[0094] In some embodiments, merging the second-format traffic light image with M channels and the traffic light image with N channels into a traffic light image to be identified with P channels includes:

[0095] Create an empty array of images;

[0096] The array data corresponding to the M channels and the array data corresponding to the N channels are sequentially written into the image array to obtain the image array with P channels;

[0097] An image of a traffic light to be identified is generated based on the image array with P channels.

[0098] In some embodiments, the first format is RAW format.

[0099] The recognition device provided in this embodiment can obtain a first-format traffic light image and a second-format traffic light image acquired by an image acquisition device; wherein the second-format traffic light image has M channels; then, the first-format traffic light image is parsed into a traffic light image with N channels; the second-format traffic light image with M channels and the traffic light image with N channels are merged into a traffic light image to be recognized with P channels; finally, the traffic lights in the traffic light image to be recognized are identified to obtain the traffic light recognition result. Since the channel colors of the P channels in the traffic light image to be recognized can reflect sufficient light information, the traffic lights can be accurately identified in the traffic light image even in nighttime scenes, improving the recognition accuracy.

[0100] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated further here.

[0101] Figure 6 This is a schematic diagram of the structure of a car shown in an embodiment of this application.

[0102] See Figure 6 In this embodiment, the vehicle 1000 includes a memory 1010 and a processor 1020.

[0103] The processor 1020 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0104] Memory 1010 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. ROM may store static data or instructions required by processor 1020 or other modules of the computer. Permanent storage devices may be read-write storage devices. Permanent storage devices may be non-volatile storage devices that retain stored instructions and data even when the computer is powered off. In some embodiments, permanent storage devices use mass storage devices (e.g., magnetic or optical disks, flash memory) as permanent storage devices. In other embodiments, permanent storage devices may be removable storage devices (e.g., floppy disks, optical drives). System memory may be a read-write storage device or a volatile read-write storage device, such as dynamic random access memory. System memory may store some or all of the instructions and data required by the processor during operation. Furthermore, memory 1010 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (e.g., DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory 1010 may include a removable storage device that is readable and / or writable, such as a laser disc (CD), a read-only digital multifunction optical disc (e.g., DVD-ROM, dual-layer DVD-ROM), a read-only Blu-ray disc, a high-density optical disc, a flash memory card (e.g., SD card, mini SD card, Micro-SD card, etc.), a magnetic floppy disk, etc. The computer-readable storage medium does not contain carrier waves or transient electronic signals transmitted wirelessly or via wire. The power battery can power various components in the electric aircraft.

[0105] The memory 1010 stores executable code, which, when processed by the processor 1020, can cause the processor 1020 to execute part or all of the methods described above.

[0106] Furthermore, the method according to this application can also be implemented as a computer program or computer program product, which includes computer program code instructions for performing some or all of the steps in the method described above.

[0107] Alternatively, this application may be implemented as a computer-readable storage medium (or a non-transitory machine-readable storage medium or a machine-readable storage medium) storing executable code (or computer program or computer instruction code) thereon, which, when executed by a processor of an electronic device (or server, etc.), causes the processor to perform part or all of the steps of the methods described above according to this application.

[0108] The various embodiments of this application have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for identifying traffic lights, characterized in that, include: A first-format traffic light image and a second-format traffic light image are acquired by an image acquisition device; wherein, the first-format traffic light image includes the original photosensitive data and shooting metadata of the image acquisition device at the time of shooting, and the second-format traffic light image has M channels; The traffic light image in the first format is parsed into a traffic light image with N channels; where M and N are both positive integers. The second-format traffic light image with M channels and the traffic light image with N channels are merged into a traffic light image to be identified with P channels; The traffic lights in the image of the traffic lights to be identified are identified to obtain the traffic light identification result.

2. The method according to claim 1, characterized in that, The acquisition of the first format traffic light image and the second format traffic light image acquired by the image acquisition device includes: The first-format traffic light image acquired by the image acquisition device is obtained based on the image acquisition code; The image acquisition device outputs a second format signal light image after converting the first format signal light image.

3. The method according to claim 1, characterized in that, The process of identifying traffic lights in the image of traffic lights to be identified, and obtaining traffic light identification results, includes: The image of the traffic light to be identified is input into the recognition neural network model to obtain the traffic light recognition result of the traffic light recognition neural network model in the image of the traffic light to be identified; The traffic light recognition neural network model is obtained by training a model based on traffic light images with P channels.

4. The method according to claim 3, characterized in that, The method further includes: Receive an image acquisition command for a target functional model that has a different function from the recognition neural network model; The traffic light image in the second format is sent to the target function model so that the target recognition model can perform the function corresponding to the target function model based on the traffic light image in the second format.

5. The method according to claim 1, characterized in that, The step of parsing the traffic light image in the first format into a traffic light image with N channels includes: The first format of the traffic light image is analyzed to obtain the arrangement of the photosensitive elements; Based on the arrangement of the photosensitive elements, the single channel of the first format signal light image is separated into N channels; Generate a signal light image with N channels based on the N channels.

6. The method according to claim 5, characterized in that, The step of merging the second-format traffic light image with M channels and the traffic light image with N channels into a traffic light image to be identified with P channels includes: Create an empty array of images; The array data corresponding to the M channels and the array data corresponding to the N channels are sequentially written into the image array to obtain the image array with P channels; An image of a traffic light to be identified is generated based on the image array with P channels.

7. The method according to claim 1, characterized in that, The first format is RAW format.

8. A traffic light identification device, characterized in that, include: The acquisition module is used to acquire a first-format traffic light image and a second-format traffic light image acquired by the image acquisition device; wherein, the first-format traffic light image includes the original photosensitive data and shooting metadata of the image acquisition device at the time of shooting, and the second-format traffic light image has M channels; The parsing module is used to parse the traffic light image in the first format into a traffic light image with N channels; wherein M and N are both positive integers; The merging module is used to merge the second-format traffic light image with M channels and the traffic light image with N channels into a traffic light image to be identified with P channels; The recognition module is used to identify the traffic lights in the image of the traffic lights to be identified and obtain the traffic light recognition result.

9. A car, characterized in that, include: processor; as well as A memory having executable code stored thereon, which, when executed by the processor, causes the processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium having executable code stored thereon, which, when executed by a processor of an electronic device, causes the processor to perform the method as described in any one of claims 1-7.

Images