Application Of Edge Preserving Filtering Technique In Space Color Gamut Mapping

In cross media image reproduction technology,the accurate transmission and reproduction of color between different media is the important issues during reproduction process,and gamut mapping technique is the key technology to effectively keep the image details,improve the color reproduction level.There are a lot of mature color gamut mapping algorithms which can be divided into two categories according to mapping theory: pointwise gamut mapping algorithm and spatial gamut mapping algorithm.The first kind of pointwise gamut mapping algorithm refers to map the color from the original color gamut to the target color gamut point by point.The advantage of this algorithm is that the mapping speed is fast and point-to-point mapping algorithm is in pursuit of absolute reduction of each color and smaller color difference after mapping.However,the color in image is different from color block,the visual perception of each color point constituting the image is not independent.Every color pixel in the image will be affected by the surrounding pixels,but the pointwise gamut mapping algorithm does not take into account the interaction between the pixels in the space,especially the adjacent pixels in the image mapping process.Therefore,it will cause the loss of image edge details in the mapping process and mapping results are not very satisfactory.The second kind of spatial gamut mapping algorithm is designed to solve the problem of pointwise gamut mapping algorithm.Spatial gamut mapping algorithm considers the characteristics of human vision and the spatial location of the image,so the relationship between pixels is protected.Compared with the traditional pointwise gamut mapping algorithm,the spatial gamut mapping algorithm based on the color relation of each mapping point is more consistent with human visual characteristic and is the more reasonable gamut mapping algorithm.The detail compensation method of the spatial gamut mapping algorithm is studied in this paper.Based on the theory and technology of spatial gamut mapping,process of color gamut mapping and typical spatial gamut mapping algorithm are researched.And the image detail extraction technology of patial gamut mapping algorithm is analyzed and compared in detail.In this paper,image filtering technique,one of the image detail extraction techniques,is firstly studied.Conventional filter,concluding mean filter and Gauss filter,is specifically analyzed.Then based on the existing mature Bala algorithm with good mapping performance,a new color gamut mapping algorithm with bilateral filter is put forward.Through the decomposition of the original image’s detaillayer and base layer,parameter setting of the bilateral filter and how color gamut mappinginfluences on the halo phenomenon are analyzed in detail.And the mapping results of the new algorithm,classical pointwise mapping algorithm and other spatial gamut mapping algorithms are compared through apsychophysicalexperiment.The new algorithm has the same mapping results as the classical compensation algorithms,and can inhibit the emergence of halo in the process of mapping more effectively.Through the research of the bilateral filter mapping algorithm,it is found that he halo problem can be effectively controlled but cannot be completely eliminated.In this paper,the guided filtering is proposed to replace the bilateral filtering for image mapping processing for the first time.Guided filtering is a fast filtering with high computational efficiency.At the same time,the performance of guided filtering gamut mapping algorithms is discussed and analyzed systematically including the effect of guided filter parameter setting on detail reproduction and the reason guided filtering eliminating halo compared with bilateral filtering.Finally through a psychophysical experiment,the image preference and mapping accuracy of the new algorithm are evaluated and the computing efficiency is tested.The experimental results show that the algorithm proposed almost has the same computing efficiency as Balasubramanian’s,but has the better performance in avoiding the emergence of halo as well as other aspects.