Research On Sequential Process Of Corrected Image Taken In Traffic Accident Scene

The number of motor vehicles has increased sharply with the rapid development of national economy. But road transport and management facilities are relatively backward, and human factors issues, it is contributes to the frequent accident happens which not only having a great impact and limitation to the development of social, but also resulting in great loss of life and property. In order to solve the delays and mishandling investigation of traffic accidents and avoid domino effect of the accidental loss, which need to find a fast and efficient instrument for surveying the scene of the traffic accident. With the deepen development of digital image photography and image progressing skills, the distance between objects at traffic accident scene can be gained through mapping and the scene's virtual map and plan view can be draw according to requirement.Because the perambulation and draft of accident scene are the basis of approving the accident process, and they are the main evidence of judging accident responsibility, so the data must be real and effective. In the image generation process, as the inherent non-linear of imaging system, and the gesture position changes of camera system and so on, which will lead to geometric distortion of images. These distortions will have serious impact on subsequent processing, so geometric correction must be done before the subsequent processing for image. The processed image looks like shooting from the top of the vertical ground, so it can accurately reflect the data of the scene.Image mosaics technology of photos which have been geometry corrected was mainly to show a complete scene of a traffic accident in the process of acquiring information. Using the sequential technology of traffic accident scene image based on geometry correction, all information covered the whole accident scene can be obtained.Based on the linear model, the process of camera imaging analyzed in details and mathematics model established for bird's-eye-photo geometry correction. The bird's-eye-photo geometry correction belongs to geometry operation of image handling technology. The image geometry operation is made up of transformation of image space coordinate and determination of pixel value. In order to geometric correction for the scene image, the paper put forwards a method-called four point method which can acquire point coordinate quickly. According to the nearest point method and pixel filling method, the process of determining pixel value can be carried out.It researches with emphasis on the two core technology of image mosaic, which are the image registration technology and the image fusion technology. The method of image matching adopted in this paper is character point matching. After all the matching points are selected, a matching line can be confirmed on each stitching image. The matching points on each image are the same points. So reference frames can be unified into two images. Furthermore, for the image fusion technology of photos which have been stitched, with weighted average method, using the gaussian template to smoothing the mosaicing image.The software was developed by using Visual C++ and based on image stitching model and image correction model. The paper described the function of each module and the operation steps.Finally, on the one hand, the paper analyzed the impact of error on geometric correction through indoor simulation experiment. The experiment was arranged according to three influence factors-the reference point region, the shoot height and the shoot angle. On the other hand, it analyzed the impact between the number of image mosaics and image accuracy with outdoor experiment, than proved the feasibility and effectiveness of the theory and algorithm in the paper.