Finding the camera displacement in a 3D measurement system through matching the background texture of stereo images using hybrid GA

For 3D measurements using image processing system, one of the most important parameters that should be determined in a single camera parallel-axis system is the precise value of camera displacement. In this system when the camera is displaced, all image background points move with the same translation vector with respect to their initial positions in the first image without any other transformation. In this paper we propose a texture matching based technique for calculating this parameter. In this approach we find the best matching between the existent background textures in the stereo images with subpixel precision. This is accomplished by subpixel registration of texture areas. The subpixel image registration is formulated as a hybrid genetic optimization problem. The used genetic algorithm has the advantages of both classical (local) and global optimization algorithms. The precision of this method is 0.01 millimeter for determining the camera displacement.