Stereo correspondence matching with Clifford phase correlation

摘要

Detecting conjugate pairs in images, known as correspondence problem is one the most challenging tasks in computer vision and image processing. In this paper we develop a phase correlation based technique to find the correspondence objects in stereo images. We use the Clifford Fourier transform to obtain phase correlation on color stereo images. With the developed Clifford phase correlation technique we can emphasize the role of color and extract accurate information from color space. We examine the accuracy and reliability of the Clifford phase correlation for stereo correspondence with color stereo images versus the normal phase correlation on gray scale images.