CN105635597B - The automatic explosion method and system of in-vehicle camera - Google Patents

Abstract

A kind of automatic explosion method and system of in-vehicle camera, are arranged area-of-interest after initialization, are all handled just for area-of-interest every frame image；Start camera, the time for exposure of first frame uses initial exposure time, it carries out that the second frame is arranged by the quick adjusting of standard of gray threshold, to follow-up each frame, present frame and previous frame image gray average difference are not above gray scale shake saltus step range and then retain Time Exposure value using present frame, else if detecting that traffic sign then carries out the exposure based on image entropy and adjusts, else if the gradation of image mean value of present frame then directly uses the value of present frame in standard grayscale threshold range, otherwise gray average difference carries out being adjusted according to the PID of gray scale and exposure value ratio if within the scope of saltus step, otherwise carry out carrying out the Time Exposure discreet value of the exposure method acquisition next frame of fast zoom table using cmos sensor characteristic；Finally discreet value is controlled according to time for exposure adjustable range.

Description

Automatic exposure method and system for vehicle-mounted camera

technical field

The invention relates to the technical field of intelligent transportation, and provides an automatic exposure method and system for a vehicle-mounted camera.

Background technique

With the improvement of living standards and the acceleration of people's life rhythm, vehicles are becoming more and more popular, and the pressure on transportation has also increased unprecedentedly. The issue of active safety of automobiles has gradually become the focus of attention.

The proposal of unmanned vehicles is based on this demand. Unmanned vehicles integrate many technologies such as automatic control, architecture, artificial intelligence, and visual computing. They are products of highly developed computer science, pattern recognition, and intelligent control technologies. It is also an important symbol to measure a country's scientific research strength and industrial level, and has broad application prospects in the fields of national defense and national economy.

A driverless car is a smart car that senses the road environment through an on-board sensor system, automatically plans a driving route, and controls the vehicle to reach a predetermined target. As the main sensing device of vehicle road environment perception system, on-board digital camera has received extensive attention.

The automatic exposure technology is a key technology in vehicle-mounted digital cameras. Different from the traditional automatic exposure method, which uses human vision as the judging standard, the automatic exposure technology in the vehicle environment enables the machine to automatically identify the target object, such as identifying traffic signs, Target objects such as traffic lights and lane lines are the purpose of automatic exposure.

The automatic exposure method of the digital camera generally has hardware realization and software realization two kinds of ways. In the early days of digital cameras, the method of hardware control was mostly used. It has the advantages of fast control speed and good effect, but has the disadvantage of high cost of hardware equipment. With the innovation of production technology and the maturity of software exposure algorithm, the same control effect can be obtained by using software control method, and there are three main methods to effectively control exposure at low cost: 1. Adjust the aperture, 2. Control the electronic shutter, 3. Adjust the zoom gain. In the environment perception of unmanned vehicles, the latter two methods are more widely used.

At present, there are three main software methods for automatic exposure control of digital cameras:

1 Automatic exposure method based on image brightness average.

2 Automatic exposure method based on image brightness histogram.

3. Automatic exposure method based on image content.

These three methods have their own advantages and disadvantages, but due to the special environment of the vehicle environment and the highly dynamic environment with variable external environment, such as backlighting, complex road environment, and water on the road, they cannot meet the program requirements well. For these reasons, the automatic exposure method of the vehicle platform requires high adaptability, high efficiency, strong real-time performance and stability.

Contents of the invention

Different from the exposure methods of cameras and other monitoring cameras, the purpose of the present invention is to start from the vehicle-based platform, mainly considering the environment in which the car is located, the vehicle conditions are complex and changeable, and accurate identification of objects is required in different locations, so that The acquired picture data can better identify objects such as lane lines, traffic signs, traffic lights, etc.; however, the complexity of the entire system, the allocated memory and processing time should not be too large; how to quickly identify objects during high-speed driving Objects, exposure procedures are the cornerstone of the entire image recognition, how to ensure the stability of the whole process is the top priority.

The technical solution of the present invention provides an automatic exposure method for a vehicle-mounted camera, comprising the following steps:

Step 1, initialization, including configuring initial exposure time, exposure time adjustment range, standard grayscale threshold range Gt, jump range Gp conforming to PID adjustment, and grayscale jump anti-shake range Gc;

Step 2, set the region of interest, and each frame of image will only be processed for the region of interest;

Step 3: Start the camera. The exposure time of the first frame adopts the initial exposure time configured in step 1. After the first frame is shot, the average gray value of the image of the first frame is obtained for the region of interest. According to the set standard gray threshold range Gt , perform a quick adjustment based on the gray threshold value, set the exposure time of the second frame according to the adjustment result, and go to step 9;

Step 4: Compare the image grayscale average value of the current frame with the image grayscale average value of the previous frame, if the difference ΔG of the grayscale average value of the two frames does not exceed the grayscale jitter jump range Gc set in step 1, then keep using The time exposure value of the current frame is used as the estimated time exposure value of the next frame, and enter step 9, otherwise, enter step 5;

Step 5, if a traffic sign is detected from the current frame image, perform exposure adjustment based on image entropy, obtain the estimated time exposure value of the next frame, and go to step 9, otherwise go to step 6;

Step 6, if the image grayscale mean value of the current frame is within the preset standard grayscale threshold range Gt in step 1, then directly use the time exposure value of the current frame as the estimated time exposure value of the next frame, and go to step 9, otherwise go to Step 7;

Step 7, further judge according to the difference ΔG between the image grayscale average value of the current frame and the image grayscale average value of the previous frame obtained in step 4, if it is within the jump range Gp preset in step 1, then proceed according to the grayscale and exposure PID adjustment of the value ratio, obtain the estimated time exposure value of the next frame, go to step 9, otherwise go to step 8;

Step 8, carry out the exposure method of using the characteristics of the CMOS sensor to quickly look up the table, obtain the estimated time exposure value of the next frame, and enter step 9;

Step 9: Control the current estimated time exposure value of the next frame according to the exposure time adjustment range in step 1. If the time exposure estimate value of the next frame is greater than the maximum value of the exposure time adjustment range, take the maximum value is the time exposure value of the next frame finally determined, if it is less than the minimum value of the exposure time adjustment range, take the minimum value as the final time exposure value of the next frame, if it is within the exposure time adjustment range, directly set the next The estimated time exposure value of a frame is the final determined time exposure value of the next frame; after controlling the camera to capture the next frame according to the final determined time exposure value, return to step 4, and use the next frame as the new current frame cycle.

Moreover, when setting the region of interest in step 2, remove the useless car chassis cover area, the top sky area, and the edges on both sides of the camera lens.

The invention provides an automatic exposure system for a vehicle-mounted camera, which is characterized in that it includes the following modules:

The initialization module is used to configure the initial exposure time, exposure time adjustment range, standard grayscale threshold range Gt, jump range Gp conforming to PID adjustment, and grayscale jump anti-shake range Gc;

The region division module is used to set the region of interest, and each frame of image will only be processed for the region of interest;

The initial startup module is used to start the camera. The exposure time of the first frame adopts the initial exposure time configured by the initialization module. After the first frame is shot, the average gray value of the first frame image is obtained for the region of interest. According to the set standard gray Threshold range Gt, perform quick adjustment based on the gray threshold as the standard, set the exposure time of the second frame according to the adjustment result, and order the control module to work;

The anti-shake adjustment module is used to compare the image grayscale mean value of the current frame with the image grayscale mean value of the previous frame, if the difference ΔG of the grayscale mean value difference between the two frames does not exceed the grayscale jitter jump set by the initialization module The range Gc then retains the time exposure value of the current frame as the estimated time exposure value of the next frame, and commands the control module to work, otherwise the image entropy adjustment module is ordered to work;

The image entropy adjustment module is used to perform exposure adjustment based on image entropy if a traffic sign is detected from the current frame image, obtain the estimated time exposure value of the next frame, and order the control module to work, otherwise order the standard judgment module to work;

A standard judging module, used to directly adopt the time exposure value of the current frame as the estimated time exposure value of the next frame if the image grayscale mean value of the current frame is within the preset standard grayscale threshold range Gt of the initialization module, and order the control module work, otherwise order the PID control adjustment module to work;

The PID control adjustment module is used to further judge according to the difference ΔG between the image grayscale average value of the current frame obtained by the anti-shake adjustment module and the image grayscale average value of the previous frame, if it is within the jump range Gp preset by the initialization module, Then perform PID adjustment according to the ratio of grayscale and exposure value, obtain the estimated time exposure value of the next frame, order the control module to work, otherwise order the CMOS sensor adjustment module to work;

The CMOS sensor adjustment module is used to perform an exposure method using the characteristics of the CMOS sensor to quickly look up the table, obtain the estimated time exposure value of the next frame, and command the control module to work;

The control module is used to control the current estimated time exposure value of the next frame according to the exposure time adjustment range in the initialization module. If the time exposure estimate value of the next frame is greater than the maximum value of the exposure time adjustment range, take The maximum value is the final time exposure value of the next frame. If it is less than the minimum value of the exposure time adjustment range, take the minimum value as the final time exposure value of the next frame. If it is within the exposure time adjustment range, set it directly The estimated value of the time exposure of the next frame is the final determined time exposure value of the next frame; after controlling the camera to capture the next frame according to the final determined time exposure value, return to order the anti-shake adjustment module to work, and set the next frame The frame is cycled through as the new current frame.

Moreover, when setting the region of interest in the region division module, the useless car chassis cover region, the top sky region, and the edges on both sides of the camera lens are removed.

The present invention synthesizes various situations that need to be considered. In order to solve the slow time-consuming scanning of the whole image, some areas can be removed by dividing the area to save scanning time. The division area is combined with the traffic sign detection module for sub-area exposure. The rapidity is the vehicle-mounted platform. One of the requirements, so in the initial adjustment, it is combined with the photosensitive characteristics of the CMOS sensor for quick adjustment, and at the same time, a gray threshold range is used for auxiliary adjustment, and the gray scale change range is added to prevent large-scale jump bands. The incoming image shakes.

Moreover, the exposure method based on the grayscale threshold adjustment is fast, but the accuracy is not high. In some special cases, the exposure value is not very ideal, but due to the rapidity requirements of the vehicle platform, this method is used. Large-scale adjustment, in special cases, the image entropy module is used for adjustment. The situation of backlight or forward light has always been the difficulty of vehicle automatic exposure. The introduction of image entropy evaluation adjustment can make more accurate exposure adjustment for the target object under the condition of backlight.

Therefore, the present invention has the following beneficial effects:

1. The exposure has strong real-time performance, and the performance can reach 20 frames per second.

2. The car camera can quickly identify and respond to the exposure value in the case of forward and reverse light conversion.

3. In some special situations, such as before and after passing through the bridge hole, the scene of severe water accumulation on the road also shows strong adaptability.

4. When identifying traffic signs, the recognition rate increases due to the addition of image entropy elements.

5. The whole process is very streamlined in terms of time complexity and space complexity, which reduces the memory occupied by the total system and the time consumed.

Description of drawings

Fig. 1 is a flowchart of an embodiment of the present invention.

Detailed ways

The technical solutions of the present invention will be described in detail below in conjunction with the drawings and embodiments.

Referring to Fig. 1, the working procedure of the automatic exposure method of the vehicle-mounted camera according to the embodiment of the present invention is as follows:

Step 1, initialization: during specific implementation, configuration files can be input, including initial exposure time, exposure time adjustment range, standard grayscale threshold range Gt, jump range Gp conforming to PID adjustment, and grayscale jump anti-shake range Gc. The PID is proportional integral derivative regulation. Those skilled in the art can preset the initial exposure time, the adjustment range of the exposure time, and the upper and lower limits of the range Gt/Gp/Gc. During specific implementation, those skilled in the art can also perform initial settings in other aspects by themselves, such as initial settings of gain, and various parameter settings of the camera, generally including closing jumbo frames, IP configuration, frame number setting, White balance modulation, the hardware is set to a fixed f-stop.

Step 2, divide the area: set the ROI (Region of Interest).

The specific functions of the vehicle-mounted camera are not all the scene-finding. The vehicle-mounted camera only needs to identify the established targets, and these targets are generally distributed in various areas of the image. Therefore, the strategy of regional exposure can be adopted, which not only reduces time-consuming, but also ensures The accuracy and efficiency of exposure adjustment are improved. When selecting the region of interest, those skilled in the art can preset the selection principle by themselves. It is recommended to remove the useless car chassis cover area, the top sky area, and the edges on both sides of the camera lens, and finally form an inverted trapezoidal area. In order to estimate the exposure value, the subsequent acquired frame images are only processed for this region of interest. During specific implementation, the area traversal function can be preset for the process call to extract the area of interest; or the user can manually select the area of interest by himself.

Step 3: Start the camera. The exposure time of the first frame is the initial exposure time configured in step 1. After the first frame is shot, the average gray value of the image of the first frame is obtained (the gray average value is only extracted for the region of interest, and the subsequent frames of other frames Similarly), according to the set standard gray-scale threshold range Gt, perform a quick adjustment based on the gray-scale threshold as a standard, set the exposure time of the second frame according to the adjustment result, and enter step 9.

For specific implementation, see: Zhao Fang, Luan Xiaoming, Sun Yue, etc. for specific implementation of fast adjustment based on the gray threshold. Exposure Estimation Method Based on Grayscale Image[J]. Applied Science and Technology, 2009. The implementation of this step also belongs to the initial preparation work, and the relevant content is omitted in Figure 1.

Step 4, anti-shake adjustment: compare the image grayscale average value of the current frame with the image grayscale average value retained in the previous frame, and the difference ΔG between the grayscale average values of the two frames does not exceed the grayscale jitter jump set in step 1 The range Gc retains the time exposure value of the current frame as the estimated time exposure value of the next frame, and skips to step 9. Otherwise, go to step 5 and start block adjustment.

Since the exposure time of the first frame is the initial exposure time directly set by the configuration, and the second frame is the result obtained according to the quick adjustment in step 3 and the control in step 9, so starting from the third frame, the image gray value of the current frame is the same as that of the previous one. For example, when step 4 is executed for the first time, the currently captured frame is the second frame, and the next frame to determine the exposure time is the third frame, and the goal is to estimate the exposure of the third frame Time value, compare the image grayscale average value of the current second frame with the image grayscale average value retained in the first frame. Subsequent return to step 4 is performed in the same way.

Different from the usual home camera camera, in the car platform, due to the changeable road environment and complex external conditions, the value of exposure adjustment is not a fixed value, but will continue to change in an area, and this change sometimes Because of a very short-term jump, such as the camera lens being blocked by a tree, the image may change from bright to dark. In order to prevent this from happening, an adjustment limit to prevent shaking has been added, so that Ensure the stability of image acquisition.

Step 5, image entropy adjustment: If a traffic sign is detected from the current frame image, perform exposure adjustment based on image entropy, obtain the estimated time exposure value of the next frame, and skip to step 9. If not detected, go to step 6.

During specific implementation, the detection of lane lines, traffic signs, traffic lights, etc. can adopt relevant prior art, and the present invention only involves the realization of exposure control. Exposure adjustment based on image entropy can be found in: Yang Zuoyan, Ruan Ping, Zhai Bo, etc. An automatic exposure algorithm for high dynamic range scenes based on image entropy[J]. Acta Photonica Sinica, 2013.

Step 6, if the image grayscale mean value of the current frame is within the preset standard grayscale threshold range Gt in step 1 (ie ) then directly retain the time exposure value of the current frame as the estimated time exposure value of the next frame, and skip to step 9. If exceeded, go to step 7.

Step 7, PID control adjustment: further judge according to the difference ΔG between the image grayscale mean value of the current frame obtained in step 4 and the image grayscale mean value value ΔG of the previous frame, if the difference is small, it meets the preset jump range Gp (ie ), that is, within the range that can be precisely adjusted, then perform the PID adjustment method based on the ratio of grayscale and exposure value to obtain the estimated time exposure value of the next frame, and skip to step 9. If it exceeds the precise adjustment range too much, go to step 8.

During the specific implementation, the specific implementation of PID adjustment according to the ratio of grayscale and exposure value can be found in: Chen Shanwei, Hou Dexin, Ye Shuliang, etc. Fast automatic exposure method in lighter flame image analysis [J]. Computer Engineering, 2011, 37(23): 200-202.

Step 8, based on CMOS sensor adjustment: because the jump between the current frame and the previous frame is too large, use the CMOS sensor characteristics to quickly look up the exposure method to achieve the effect of quickly locking the exposure range, and obtain the time exposure preview of the next frame. Valuation, go to step 9.

During specific implementation, the specific implementation of the exposure method of using the characteristics of CMOS sensors to quickly look up the table can be found in: Vuong Q K, Yun S H, Kim S.A new auto exposure system to detect high dynamic range conditions using CMOS technology [C]. 2008 Third International Conference on Convergence and Hybrid Information Technology, Busan 2008:577-580.

Step 9, exposure threshold range control: control the estimated time exposure value of the next frame currently obtained according to the exposure time adjustment range in step 1, if the estimated time exposure value of the next frame is greater than the maximum value of the exposure time adjustment range , then take the maximum value as the final time exposure value of the next frame, if it is less than the minimum value of the exposure time adjustment range, then take the minimum value as the final time exposure value of the next frame, if it is within the exposure time adjustment range Then directly set the time exposure estimate value of the next frame as the final determined time exposure value of the next frame, and then control the camera to capture the next frame, and then return to step 4 to start taking the next frame as the new current frame Loop, which can run automatically until the user asks to stop.

In order to prove the effect of the present invention, each method can be tested, and the effect can be judged according to the time-consuming and frame rate. The test table is as follows:

Each method test is based on the time-consuming and frame rate to judge the effect. The test table is as follows:

It can be seen that the method provided by the present invention has a significant effect of improving the recognition rate, and the time consumption will not be prolonged.

During specific implementation, the method provided by the present invention can use software technology to realize automatic operation, and can also provide a corresponding system in a modular manner. The embodiment of the present invention also provides an automatic exposure system for a vehicle-mounted camera, including the following modules:

The initialization module is used to configure the initial exposure time, exposure time adjustment range, standard grayscale threshold range Gt, jump range Gp conforming to PID adjustment, and grayscale jump anti-shake range Gc;

The region division module is used to set the region of interest, and each frame of image will only be processed for the region of interest;

The initial startup module is used to start the camera. The exposure time of the first frame adopts the initial exposure time configured by the initialization module. After the first frame is shot, the average gray value of the first frame image is obtained for the region of interest. According to the set standard gray Threshold range Gt, perform quick adjustment based on the gray threshold as the standard, set the exposure time of the second frame according to the adjustment result, and order the control module to work;

The anti-shake adjustment module is used to compare the image grayscale mean value of the current frame with the image grayscale mean value of the previous frame, if the difference ΔG of the grayscale mean value difference between the two frames does not exceed the grayscale jitter jump set by the initialization module The range Gc then retains the time exposure value of the current frame as the estimated time exposure value of the next frame, and commands the control module to work, otherwise the image entropy adjustment module is ordered to work;

The image entropy adjustment module is used to perform exposure adjustment based on image entropy if a traffic sign is detected from the current frame image, obtain the estimated time exposure value of the next frame, and order the control module to work, otherwise order the standard judgment module to work;

A standard judging module, used to directly adopt the time exposure value of the current frame as the estimated time exposure value of the next frame if the image grayscale mean value of the current frame is within the preset standard grayscale threshold range Gt of the initialization module, and order the control module work, otherwise order the PID control adjustment module to work;

The PID control adjustment module is used to further judge according to the difference ΔG between the image grayscale average value of the current frame obtained by the anti-shake adjustment module and the image grayscale average value of the previous frame, if it is within the jump range Gp preset by the initialization module, Then perform PID adjustment according to the ratio of grayscale and exposure value, obtain the estimated time exposure value of the next frame, order the control module to work, otherwise order the CMOS sensor adjustment module to work;

The CMOS sensor adjustment module is used to perform an exposure method using the characteristics of the CMOS sensor to quickly look up the table, obtain the estimated time exposure value of the next frame, and command the control module to work;

The control module is used to control the current estimated time exposure value of the next frame according to the exposure time adjustment range in the initialization module. If the time exposure estimate value of the next frame is greater than the maximum value of the exposure time adjustment range, take The maximum value is the final time exposure value of the next frame. If it is less than the minimum value of the exposure time adjustment range, take the minimum value as the final time exposure value of the next frame. If it is within the exposure time adjustment range, set it directly The estimated value of the time exposure of the next frame is the final determined time exposure value of the next frame; after controlling the camera to capture the next frame according to the final determined time exposure value, return to order the anti-shake adjustment module to work, and set the next frame The frame is cycled through as the new current frame.

Further, when the region of interest is set in the region division module, the useless car chassis cover region, the top sky region and the edges on both sides of the camera lens are removed.

The above embodiments are only for the purpose of illustrating the present invention, rather than limiting the present invention. Those skilled in the relevant technical fields can also make various changes or modifications without departing from the spirit and scope of the present invention. Therefore, all Equivalent technical solutions all fall within the protection scope of the present invention.

Claims (4)

1. an automatic exposure method for a vehicle-mounted camera, comprising the following steps: Step 1, initialization, including configuring initial exposure time, exposure time adjustment range, standard grayscale threshold range Gt, jump range Gp conforming to PID adjustment, and grayscale jump anti-shake range Gc; Step 2, set the region of interest, and each frame of image will only be processed for the region of interest; Step 3: Start the camera. The exposure time of the first frame adopts the initial exposure time configured in step 1. After the first frame is shot, the average gray value of the image of the first frame is obtained for the region of interest. According to the set standard gray threshold range Gt , perform a quick adjustment based on the gray threshold value, set the exposure time of the second frame according to the adjustment result, and go to step 9; Step 4: Compare the image grayscale average value of the current frame with the image grayscale average value of the previous frame, if the difference ΔG of the grayscale average value of the two frames does not exceed the grayscale jump anti-shake range Gc set in step 1, then Keep the time exposure value of the current frame as the estimated time exposure value of the next frame, go to step 9, otherwise go to step 5; Step 5, if a traffic sign is detected from the current frame image, perform exposure adjustment based on image entropy, obtain the estimated time exposure value of the next frame, and go to step 9, otherwise go to step 6; Step 6, if the image grayscale mean value of the current frame is within the preset standard grayscale threshold range Gt in step 1, then directly use the time exposure value of the current frame as the estimated time exposure value of the next frame, and go to step 9, otherwise go to Step 7; Step 7, further judge according to the difference ΔG between the image grayscale average value of the current frame and the image grayscale average value of the previous frame obtained in step 4, if it is within the jump range Gp preset in step 1, then proceed according to the grayscale and exposure PID adjustment of the value ratio, obtain the estimated time exposure value of the next frame, go to step 9, otherwise go to step 8; Step 8, carry out the exposure method of using the characteristics of the CMOS sensor to quickly look up the table, obtain the estimated time exposure value of the next frame, and enter step 9; Step 9: Control the current estimated time exposure value of the next frame according to the exposure time adjustment range in step 1. If the time exposure estimate value of the next frame is greater than the maximum value of the exposure time adjustment range, take the maximum value is the time exposure value of the next frame finally determined, if it is less than the minimum value of the exposure time adjustment range, take the minimum value as the final time exposure value of the next frame, if it is within the exposure time adjustment range, directly set the next The estimated time exposure value of a frame is the final determined time exposure value of the next frame; after controlling the camera to capture the next frame according to the final determined time exposure value, return to step 4, and use the next frame as the new current frame cycle. 2. The automatic exposure method of the vehicle-mounted camera according to claim 1, characterized in that: when the region of interest is set in step 2, the useless car chassis cover area, top sky area and both sides of the camera lens are removed. 3. An automatic exposure system for a vehicle-mounted camera, comprising the following modules: The initialization module is used to configure the initial exposure time, exposure time adjustment range, standard grayscale threshold range Gt, jump range Gp conforming to PID adjustment, and grayscale jump anti-shake range Gc; The region division module is used to set the region of interest, and each frame of image will only be processed for the region of interest; The initial startup module is used to start the camera. The exposure time of the first frame adopts the initial exposure time configured by the initialization module. After the first frame is shot, the average gray value of the first frame image is obtained for the region of interest. According to the set standard gray Threshold range Gt, perform quick adjustment based on the gray threshold as the standard, set the exposure time of the second frame according to the adjustment result, and order the control module to work; The anti-shake adjustment module is used to compare the image gray level mean value of the current frame with the image gray level mean value of the previous frame. The jitter range Gc then retains the time exposure value of the current frame as the estimated time exposure value of the next frame, and the command control module works, otherwise the image entropy adjustment module is commanded to work; The image entropy adjustment module is used to perform exposure adjustment based on image entropy if a traffic sign is detected from the current frame image, obtain the estimated time exposure value of the next frame, and order the control module to work, otherwise order the standard judgment module to work; A standard judging module, used to directly adopt the time exposure value of the current frame as the estimated time exposure value of the next frame if the image grayscale mean value of the current frame is within the preset standard grayscale threshold range Gt of the initialization module, and order the control module work, otherwise order the PID control adjustment module to work; The PID control adjustment module is used to further judge according to the difference ΔG between the image grayscale average value of the current frame obtained by the anti-shake adjustment module and the image grayscale average value of the previous frame, if it is within the jump range Gp preset by the initialization module, Then perform PID adjustment according to the ratio of grayscale and exposure value, obtain the estimated time exposure value of the next frame, order the control module to work, otherwise order the CMOS sensor adjustment module to work; The CMOS sensor adjustment module is used to perform an exposure method using the characteristics of the CMOS sensor to quickly look up the table, obtain the estimated time exposure value of the next frame, and command the control module to work; The control module is used to control the current estimated time exposure value of the next frame according to the exposure time adjustment range in the initialization module. If the time exposure estimate value of the next frame is greater than the maximum value of the exposure time adjustment range, take The maximum value is the final time exposure value of the next frame. If it is less than the minimum value of the exposure time adjustment range, take the minimum value as the final time exposure value of the next frame. If it is within the exposure time adjustment range, set it directly The estimated value of the time exposure of the next frame is the final determined time exposure value of the next frame; after controlling the camera to capture the next frame according to the final determined time exposure value, return to order the anti-shake adjustment module to work, and set the next frame The frame is cycled through as the new current frame. 4. The automatic exposure system of the vehicle-mounted camera according to claim 3, characterized in that: when the region of interest is set in the region division module, the useless car chassis cover region, top sky region and both sides of the camera lens are removed.