Interactive Image Editing Technique Based On Resample

Nowadays,with the development and the progress of science and technology,consequent camera equipment has been gradually generalized,thus a large number of pictures springs up every day,and the image have become an important media for transmitting information. Followed by people have a higher demand of the quality of images. However, when photos taken directly couldn't meet people's requirement, image editing becomes an important supplementary means at this time. Take the following two problems as an example. One is that some certain areas of the image tone slants dark or brightness, or some certain colors need to be adjusted. The other is that the background of the image is inappropriate and needs to be replaced. These two types of problem can be solved by separately using color editing and image matting technology, and this article has presented an interactive editing technology based on resampling.Interactive image editing commonly uses brushes to choose editing reference sample and provide appropriate editing mode. However, the editing effects of existing methods is too dependent on the choice of input samples, in order to get the desired editing effect, it usually requires specially to select a representative and more abundant sample. To simplify the input to reduce the burden on the operator, this article has studied an efficient resampling edit propagation technique. First, using super-pixel segmentation technique, based on pixel color similar principle within each super-pixel, choose the pixel whose color value close to pixel color average within super-pixel as the resampling sample. Then comply with the similarity between resampling sample and input sample and the similarity of resampling sample to edit and propagate the input sample to the resampling sample. Finally, edit and propagate the resampling sample to the whole image of all pixels. However, using color information rich resampling samples to replace brush selected input reference samples to propagate, which not only can simplify user interaction, reduce the operator's requirements, but also can reduce the calculation amount of data, thereby enhancing the efficiency of the program.This paper proposes the resampling editing technology, which are separately used for two image editing problems of color transfer and image matting. When performing color transfer, adopt radial basis function interpolation(RBF) to propagate the new color of the resampling sample to all the pixels to make sure that abides by input reference editing mode and keep the details of the images at the same time. When performing image matting, use matting Laplacian to establish optimization function to edit and propagate the resampling sample to neighboring pixels in accordance with the principle of local similarity. The results can be found by the experiment, in the case of providing the simple input reference sample, resampling editing propagation technology presented in this article is better than direct propagation in the effect. What's more, the image matting algorithms presented herein does not require the aid of FIG thirds(trimap), which greatly simplifies the user input.