A novel approach for approving constant image brightness

Stereo vision is a technique used to recover depth from a stereo pair, which can be useful in many different domains (cartography, robotics, aerial navigation). A large amount of methods take as basic assumption that intensity of corresponding points in both images are equal. But the assumption is generally false, for the possible changes of luminance between the images of the stereo pair. Consequently, if we want to recover depth from a stereo pair precisely, it is useful to give them the same dynamic range and luminance first. This paper presents a novel algorithm based on adaptive midway image equalization (A-MIE). This proposed algorithm extends on the midway image equalization studied by Delon (2004). Our method allows every block of each image to adapt to its neighboring region, so that high contrast based on human visual system can be obtained for all locations and can modify the histogram of the image pair the same dynamic range. Both theoretical analysis and experimental results show that image histograms are closely matched after A-MIE, namely its effect on the support of the Fourier spectrum demonstrate the proposed algorithm is effective.