Research On Auto Exposure Method For On-Board Digital Camera

With the improvement of living standards, people’s demands for cars are increasing sharply. So are the demands for their safety performance. Recently years, the investigation on Active Safety Technology has made considerable developments. Researchers tremendously focus the on-board digital camera which is the main sensing device of the perceptual system of road vehicles environment. Nevertheless, due to its special application background, on-board digital camera is not only compatible to the electricity and the vehicle environment, but also should have more light adaptability and more stable image output quality than ordinary cameras, the thesis mainly discusses the automatic exposure methods of the vehicle visual system and presents the mean exposure method of histogram based on the target of the on-board digital camera.The main work of this thesis is as follows:1) Owing to the low convergent speed and vibration of the traditional gray exposure method, PID control method is used in the exposure control, and a certain area is selected as area of interest. The proposed method can effectively improve the convergence rate of the traditional algorism while reducing the vibration.2) A power-based exposure estimation method is put forward, which uses the variation of the image luminance as the error factors to get the modifier of the exposure time. With analysis of shape features of V component histogram in different exposure type images, area ratio of two parts in histogram and its peak are used to identify the exposure type.3) The pixel values of changed image-RGB are used to increase the main color regions (red, blue and yellow) of the traffic signs, thus regions of interest of traffic signs are segmented. The noise is eliminated through morphologic operation. Mark figures in regions of interest are extracted to be the shape features, which are the reference quantities to sort the traffic signs with the nearest classification algorithm.4) For the detection of traffic histogram and target-oriented automatic exposure method using brightness histogram and average gray value is proposed. This method, whose target is the traffic sign of the image, makes the area of traffic sign the region of interest and then judges the exposure conditions of the whole image and region of interest by exposure estimation method based on energy. Simultaneously, according to the weights table, the mean exposure method chooses the proper weighted value and finishes the camera’s automatic exposure by combining the computing method of exposure time to obtain the most appropriate exposure time of the next image. Experimental results show that the exposured images with this method have reasonable higher recognition rate for traffic signs.