Image Shaking Calibration Algorithm for The Vehicle Image-based Surveillance System

In this paper, we proposed an image shaking calibration algorithm for the vehicle image based surveillance system. The proposed algorithm uses Vertical Edge Detection to select the confidence region of extracted image and Lucas-Canada Optical Flow technique to estimate motion. And we implemented the image correction algorithm for the selected region in the difference image between two frames in the digital image. In order to verify the efficiency of the proposed algorithm, the extracted video image was divided into 8×6 format so as to have a size of 160 pixels in width × height in 1280 × 960 size. As to the binarization result, the white boundary frequency was obtained for each area in the state of the monochrome image. Thereafter, the simulation was performed by selecting the region as the confidence region when it is above a certain threshold value. As a result, a similar restoration angle of about 85% was obtained in the frame between the original image and the difference image.