Design And Implementation Of Color Transfer Algorithms Based On Moment Features

Color transfer is one of the hottest areas in Computer Vision and Image Processing fields.This dissertation discusses and implements the color transfer algorithm wich moment feature. Some qualitative conclusion and strategy of improvement are put forward, basing on massively tests and analyzes. Result of algorithm adjusted is given, basing on the implementation of two Lαβ color space transfer algorithms, which are with logarithm and without logarithm. By testing, the algorithm without logarithm is chosen for following work. Three color transfer algorithm with lower moments are implemented and application environment is given basing massively tests. In the aspect of chosing of source images and target images, some qualitative and practical conclusion is given.Basing above work, higher moment is introduced into matching process and strategy of improvement is put forward, considering low precision of low moment in matching process. Result of tests indicates that the algorithm improves the matching precision and effect of color transfer. Basing analyzing factors which affects the matching efficiency, a improved matchming strategy with filtering is put forward. Result of tests indicates that the strategy improves the algorithm efficiency by fitering massive irrelevant samples. The algorithms discussed in this dissertation have the widespread application prospect in many aspects, for instance, coloring for old gray photoes, gray movie piece,gray medicine picture, gray satellite picture and so on, the color correcting for color image, and special artistic effect.