A Progressive Color Transfer Algorithm And Its Applications

In recent years,colorization technique is a challenging research area in the hot realm of digital image processing.Image color transfer technique is a process that transfers the specific color mode to the specified image.By analyzing the ideas and features of classic colorization techniques,this paper presents and realizes a progressive color transfer of controllable principal object.The main works in this dissertation are as follows:1)By summarizing and generalizing the features and principles of several current colorization algorithms,this paper describes the idea of the algorithm in detail.It tests and analysis the experimental results to obtain a number of qualitative conclusions of these algorithm.2)One problem with the point matching-based color transfer method is the possibly erroneous transfer of the color of principal objects.To deal with this problem,a color transfer algorithm with controllable principal objects is proposed.The algorithm first separates the principal objects from the background by building a mask of principal objects via image segmentation and carries out shape analysis of histograms of both source and target images,then figures out zeros of histograms using forward differential computation and thereby identifies the structure of peaks and valleys of both histograms by taking into account the information of second order derivatives,and finally completes the progressive correct color transfer of the principal objects through gradual histogram peaks matching of the objects.Experiment results with the algorithm shows a good control over principal object color transfer without much manual intervention.3)We achieved this color transfer algorithm with controllable principal objects on Matlab platform and generated the corresponding software framework.Through this interactive user interface,we can control the source image,the target image and the mask,it is convenient for users debugging and getting the best color transfer result.The old black-white photo and video sequences only have luminance information,and the cognitive ability of human eye to brightness is far worse than to color,colorization can enhance the image information,and increase the visual effect.Coloring the black-white photo and classic movie can improve the visual delight.In this paper we use progressive color histogram matching to control the color transfer of the image principal object.The method is simple and effective,does not require too much user intervention.We just formulate a matching degree,it can automatically generate the final output.Since the principal object of the image often has rich colors,some of the traditional method is relatively difficult to transfer the color,we can solve this problem by the method of this paper and get some good results.Meanwhile,we can also be free to control the transfer target,or full map transfer or partial transfer,so that the whole process may be more flexible,and can be able to select the desired intermediate process as the final result.