Application of coherency sensitive hashing algorithm in image analogies

Image analogies algorithm proposed by Hertzmann uses the pixel search methods of approximate nearest neighbor search (ANN) and the coherence search. The algorithm of image analogies works well in image effects, but is slow to synthesize. In this paper, we use coherency sensitive hashing (CSH) algorithm to improve the searching and matching speed of the ANN algorithm in the premise of not reducing the synthesis effects of images. This algorithm increases the practicability of the image analogies.