Improvement And Implementation Of Interactive Graph Cut Algorithm

In recent years, digital image analysis and processing technology has been more and more attention from researchers and enterprises. Although the development history of the technology is not too long but because its processing technology and theory algorithm variety, phyletic and various, so it has gradually become the generally accepted one of the most development potential of the science project.Known, physical human recognition or accept to all kinds of information, mainly includes four categories they are static images, voice, text, dynamic images. And studies show static images in these four categories in the brain is the most profound memory. The vast majority of information that is stored in the brain should be images.That is why people are so interested in image technology. This paper briefly describes the image of cutting it is very important a preprocessing step in the process of image recognition technology, it is a very important part of image technology. So I think the research of image cutting technology is of great significance and practical.The application scene and the application range of image cutting technology is very wide. You can say that if anyone comes to want to belong to the image of target image cutting part cutting technology is inseparable from the images. Its importance is more outstanding especially in the medical field., chest pictures such as heart, brain images, and cut cell images are the concrete embodiment of image cutting technology in the medical application. Bovis used texture and fragmentation region merging method of X-ray image cutting, and the threshold of cutting rule is used by CT scanning. Microsoft Corp produced a body sensing device in Kinect Albiold on the use of preprocessing method for video surveillance. Tan proposed the normalized illumination feature and local texture successfully solves the effect of natural light on the face recognition. In the application of machine learning field image cuttingtechnology is quite extensive. And image technology, image information transmission of mobile technology and equipment in the development of application and so on.Interactive graphics cutting algorithm suggests that when cutting graphics users need to participate in, there are some user specified cutting conditions, to help complete the task of cutting. A large number of applications to the field of graphic apply interactive graphics cutting algorithm. In contrast, non interactive graphics cutting algorithm does not require user participation, the algorithm will automatically complete graphics cutting work. Target recognition and visual field is the main application of this kind of algorithm. In fact, interactive graphics cutting algorithm and non interactive graphics cutting algorithm have not essential difference, only different cutting conditions.think the interactive graph cut algorithm is the best and the most practical in many graph cut algorithm, the characteristic of the algorithm is a complex problem into the problem of cutting image pixel images labeled as foreground or background pixels. First,you must regard the picture as a network diagram, after take the initiative to designate specified number of pixels as seed pixels, other pixels naturally is the network nodes in the graph. Two adjacent pixels you can be regarded as a network graph exists edge. Based on the existing network nodes add two additional nodes you can construct a S-T diagram. The maximum flow minimum cut algorithm to obtain the desired segmentation result. Now the mobile terminal software is very popular,realize the graph cut algorithm has great practical value in the mobile terminal.This paper in the original graph cut algorithm for improving the algorithm, first discusses the maximum flow minimum cut algorithm, and the related knowledge of network flow theory to the improved algorithm after a theory. Interactive graph cut algorithm because of the impact of the regional items unable to accurately obtains wants cutting results, so the author will choose a region into the boundary, eliminate the influence brought by region. Finally, the algorithm is implemented on the Android platform and the experimental data are tested to prove the superiority of the algorithm.