Color Correction Based on SIFT and GRNN for Multi-view Video

In order to correct color differences between different views, a color correction method based on Scale Invariant Feature Transform (SIFT) and Generalized Regression Neural Network (GRNN) is proposed. Image features of the target image and source image are firstly extracted by SIFT algorithm. The value of these corresponding image points is used to construct the GRNN neural network that corrects the image color. Experimental results show that the proposed method can produce better correction result than histogram match. It also shows that the color differences between multi-view video can be effectively solved by GRNN network.