MSHS: The mean-standard deviation curve matching algorithm in HSV space

Aiming at the loss of color information within existing curve matching methods, which happens on transformation from an RGB color image into gray space and results in mismatches, we present a novel algorithm based on the mean-standard deviation in HSV color space. This algorithm combines the mean-standard deviation with the hue-saturation color information, which is constructed by the following steps: (1) For each pixel on the feature curve, the hue and saturation information of neighbor support region are extracted respectively to compute the mean-standard deviation of the hue and saturation (MSHS) in each sub-region, which forms a four-dimensional description vector. (2) Construct the description matrix by stacking description vectors of all sub-regions associated with the curve. (3) Calculate the mean and the standard deviation vectors of description matrix and then normalize them separately. Experiments show that the proposed algorithm has high accuracy for colorful and diverse images. Moreover, the matching time is short.