One-dimension angle-oriented projection template matching

Refer to two key stages in template matching between two overlapping images: template selection and iterative matching, an one-dimension angle-oriented projection template matching method was proposed. The matching process firstly evolved a template selection discipline based on gray level difference segmentation to reduce the sensibility to image noise and illumination changes. After the segmentation, a well post intensity projection into angle-oriented array by linear interpolation was used to accelerate matching process and to maximum the employment of edge information. The method at last implemented an angle-oriented reformed AD (absolute differences) matching principle both to reduce miss-matches conduced by viewpoint rotation and to obtain rotation between reference and target other than correspondences coordinates. With aid of analysis of matching process respect to different cases under consideration of speed and precision, experimental results proved that the proposed approach works well over most overlapping images and is fast enough for real time applications.