Binocular stereo vision method based on wavelet multi-resolution and local entropy

Image matching has become a very important technology in the field of image information processing and has been widely used in photogrammetry, virtual reality scenes and biomedicine. Traditional matching algorithm based on gray-scale has neither noise immunity nor more timesaving property. In this paper, we represent a new matching algorithm which combines wavelet multi-resolution and local entropy. The identical points were matched according to the calculation of the local entropy between images to improve the ability of noise-resistance. Meanwhile, matching points of two images are established through the choosing strategy that is set from the rough to the fine by the wavelet multi-resolution to enhancing the computing efficiency. The algorithm has obvious superiorities in matching accuracy and computation for noisy images shown in the experiments finished by MATLAB.