Research On Technology Of Digital Image Inpainting About Rock

Rock paintings record the historical and cultural life of ancient human, which bear the huge ancient civilization of mankind. Rock paintings have great historic value, and is helpful for us to deeply explore the ancient culture. The techniques of digital image inpainting is the research focus in the field of digital image processing. It has important guiding significance and application value in the restoration of the damaged image, reconstruction of the cultural relics and the elimination of unwanted things in the objects. Therefore, it has great meanings to repair and protect digital rock paintings by adopting image processing technology.(1)Collect the rock paintings images, use high resolution camera to take pictures of damaged rock paintings and save as high pixel photos, which can not only accurately record the rock paintings colors, contents and forms, but also can keep high definition of the damaged areas. This method makes the restoration results more accurate, so it has a very wide applicability and high collection efficiency.(2)The basic method of image denoising is studied. In this paper, the realization of the median filter and mean filter is introduced, and the two algorithms are simulated. The simulation found they will filter the details of the image when denoising and make the images fuzzy. According to the deficiency of the above two kinds of denoising methods, In this paper, a median filter and TV denoising optimal denoising algorithm is proposed and gives a reasonable objective evaluation index values. Experimental verification shows that this innovative method can effectively suppress the edge noises and keep the details of the texture not changed.(3)For the damage of structural information, the extraction and restoration of damaged areasare carried out. The cracks are detected and extracted by a full automatic method, and according to the characteristics of rock paintings, choose the TV restoration algorithms with stronger robustness to conduct image restoration which reaches a good repair effect. The performance of the algorithm is analyzed by simulation, and it is pointed out that it has the limitation, that is, it is only meet to restore the structural images.(4)An improved Criminisi restoration algorithm is proposed. This method is based on the study of the exemplar patches restoration method, and the search method of the Criminisi algorithm to search the optimal matching block is a global search algorithm which is not only time consuming,but also is easy to produce the extension of the information and wrong block matching. A new algorithm for digital image processing algorithm inpainting is proposed by studying the characteristics of rock paintings. This algorithm improves the form of preference function of exemplar patches, and changes the searching method of the optimal block. The p_laplace operator is used to constrain the data items in the representation of the preference. In the search for the optimal matching block, it is noted that the sample blocks contain not only the color information but also the structure information. Based on this, the improvement is made and the optimization algorithm is simulated by Matlab, and the rationality and feasibility of the optimization algorithm are illustrated by the peak signal to noise ratio.