A fast image matching research based on MIC-SURF algorithm

For making up having no rotational invariance, missing many corners and susceptible to noise of the MIC algorithm, a fast image matching based on MIC-SURF algorithm is proposed which is combined with the SURF algorithm and the improved MIC algorithm. First of all, the image for adaptive smooth filtered, using neighborhood pixel similarity method to calculate the numbers of similar gray value between 8 neighborhoods of the current pixel and its nuclear; then adopting multi-grid algorithm detects geometric corners in the non-flat areas of the image; finally utilizing the SURF algorithm describes and matches feature points. Experiment results indicate that it calculates accurately and fast after the image illumination, rotation, affine and noise changing.