System Of Image And Video Color Transfer Algorithm Based On Similarity Measure

Image color migration is the process which put view’s color transfer target to the image scene. We show the image presentation which could make the image more beautiful. In our daily life, quickly and intuitive editing of digital images and video content, has become more and more important. In this paper, based on the similarity of deep color transfer algorithm, we solve the efficient and natural image processing and video color, and the brightness and contrast of the transformation problem, and implementation of a video image and color editing system.First, we introduce a similarity measure which based on image color transfer method which mainly by editing the image with complex data sets and color texture characters, to achieve the image color transfer algorithm. In the design of the algorithm framework, users can select to implement color migration area, setting the corresponding color effect and transport parameters, by adjusting the input parameters and can improve color migration effectively. In the design of the algorithm framework, users can choose to implement color transfer region, setting the corresponding color effect and transport parameters. Then they can adjust the input parameters to improve the color migration effectively. In the realization of color transfer algorithms, mainly by performing an editing propagation formula, based on the similarity of the images of characters, similar data sets as well as user input image transfer parameters, through the matrix of the vector product and similarity approximation of the equations of optimal solution, finally will obtain the color editing parameter edit propagation equation which the realization of image color transfer.Secondly, we used the GPU CUDA framework for parallel computation and high speed floating point calculation characteristics. The similarity measure based on color transfer algorithm applied to the video color migration. Through traditional Mean Shift algorithm, we using the target color histogram as a search feature, we iterative Meanshift vector makes the algorithm converges to the targets true position, to realize the video color migration regional tracking. In the movement of camera case, also can be to complete the color transfer. In the GPU acceleration and optimization of processing, mainly is that the color transfer operation in the process of color transfer algorithms in the color space conversion, feature measurement solution and parameter editing solution, the three process to the GPU for execution, on the algorithm is time consuming features of the measure solving and edit the parameters for parallel solving the acceleration and optimization. To avoid the traditional CPU operations in a serial loop handles, we obtain better color transfer effect at the same time, significantly improves the algorithm calculation speed.Finally, we introduce an image and video color edit system design and implementation process, mainly from the following aspects. The system needs analysis, system design, system development platform is introduced, the realization of the color transfer function of each module, system flow chart, system realization and system class diagram the user interface are introduced, the realization of image color editing specific operational processes and video edit color of the specific operation process. The system is based on the similarity of the color transfer algorithm as the core, the algorithm is applied to the specific image and video color editing system, test the feasibility of the proposed approach.