The Study On Methods Of Region Counting, Image Recurring And Image Rotation

Region counting is a conception in computer graphics and image analysis, and it has many applications in medical area recently. In this paper, a region counting algorithm based on region labeling automaton and an improved contour-tracing algorithm for region counting are proposed. By tracing the boundaries of the regions and the whole image once, the number of the regions can be obtained fast without filling the regions. The former algorithm can obtain the vertex chain codes when counting the regions with some additional program. And the latter one can improved the efficiency of counting region with virtual expanding border. The optimized algorithm also can get the Freeman chain codes when counting the regions, which is useful for recurring an image. And the proposed methods were found to be fastest and requiring less memory.In the researching area of the computer graphics and image, image's recurring has important practical applications. In many places where large amounts of images are used, there is a need to reduce the space needed to store the images. In this article, an image recurring algorithm based on Freeman chain code is proposed. The proposed algorithm uses the contour tracing technique. When tracing the boundaries in an image, the beginning points of the boudaries, types of them and the Freeman chain codes can be obtained. With the type of the chains, the value of filling direction can be got. These values can avoid the refilling of a multiply-connected region and improved the speed of recurring an image.Image rotation is a key link of the digital image processing and it is also a technique of image's geometry transform. In this paper, we study the existing algorithms and softwares of image rotation; present a rotation algorithm based on area interpolation for binary image. According to calculation of the four vertexes' coordinates of a pixel, the rotated coordinates of the four points can be obtained, and we can determine the possible type of the rotated pixel pane. By calculating the area in each pixel that the rotated pixel mantled, we realize the object image's interpolation and improve the quality of image rotation. This method solves the sawtooth problem and smoothes the boundaries.