Study On NAM Image Representation Methods Based On Gouraud Shading Approach And Multi-Subpattern

Image representation is an important issue in computer graphics, computer vision, robotics, image processing and pattern recognition. Efficient image representations can save space and facilitate the manipulation of the acquired images. With the advent of the digital information era and the development of the computer multimedia technology, all kinds of image data are dramatically increasing. Due to mass data of image information, the fact that the demands of the fast and real-time transmission for images have failed to meet the development of the Internet has become a difficult problem. Therefore, the study on image representation method becomes increasingly important, and it is one of the hottest research fields at present. The Non-symmetry and Anti-packing Model (NAM) is suitable for representations of image pattern, audio pattern, video pattern, and text pattern, and it is a general pattern representation model.With respect to the lossy image representation, by extending the famous Gouraud shading method, a novel gray image representation using the NAM and shading approach called the NAM Coding (NAMC) approach is proposed and is compared with the popular S-Tree Coding (STC) approach. The encoding can be performed in O(n log n) time and the decoding can be performed in O(n) time, where n denotes the number of pixels in the gray images. By taking the idiomatic standard gray image 'Lena' in the field of image processing as a typical test object, the theoretical and experimental results show that the proposed NAMC approach can reduce the bits per pixel and the number of the homogeneous blocks in the gray images much more effective than the popular STC approach and the traded JPEG approach on the premise of remaining the image quality, and therefore it is a better method to represent the lossy image pattern.Not only can the NAMC approach save storage space, but also it can facilitate the image manipulations. Some image manipulations, such as neighbor finding, searching, area computing, and set operations, are often used in some complex algorithms in the field of image processing. As a representative of applications, a fast algorithm for computing the lower order moments is proposed on the NAMC representation directly, which takes O(N) time where N is the number of NAMC blocks. By taking three idiomatic standard gray images 'Lena', 'F16', and 'Peppers' in the field of image processing as typical test objects, and by comparing the proposed NAMC approach for computing lower order moments with the popular STC approach, the theoretical and experimental results show that the former is faster than the latter with respect to the computing speed.With respect to the lossless image representation, a novel NAM image representation method based on the multi-subpattern is proposed. Also, an optimization strategy of the NAM encoder using raster scanning is presented. By taking a typical multi-subpattern (the combination of the triangle and the rectangle subpatterns) as an example, a novel image representation method by using the Non-symmetry and Anti-packing Model with Triangles and Rectangles (NAMTR) is proposed. Also, Some concrete algorithms of the NAMTR representation for binary, gray, and color images are proposed and the storage structures, the total data amount, and the time and space complexities of these proposed algorithms are analyzed in detail. As far as multi-valued images, i.e., the gray and color images, are concerned, two kinds of representation methods are put forward. One is the direct representation of the NAMTR, and the other is indirect representation of the NAMTR, i.e., the representation of the Binary-bit Plane Decomposition (BPD). The BPD method can effectively reduce the complexity of multi-valued images and achieve high representation efficiency. The total data amount and the compression ratios of the two representation methods for multi-valued images have relations with the image complexity. The lower the image complexity is, the higher the representation efficiency of the NAMTR is. By taking some idiomatic standard images, such as 'Lena', 'F16', and 'Peppers', in the field of image processing as some typical test objects, and comparing the algorithm of NAMTR with those of the classic linear quadtree (LQT), the latest NAM, and the popular compact quadtree (CQT), the theoretical and experimental results show that the former is obviously superior to the latters with respect to the numbers of subpatterns (nodes) and the data storage, and therefore it is a better method to represent the lossless image pattern.Not only can the NAMTR approach save storage space, but also it can facilitate the image manipulations. As a representative image manipulation, a fast algorithm for computing the area of a binary image is proposed on the NAMTR representation directly, which takes O(N) time where N is the total subpattern number. By taking some idiomatic standard binary images, such as 'Lena', 'F16', and 'Peppers', in the field of image processing as some typical test objects, and by comparing the proposed NAMTR approach for computing the area of the binary image with the popular CQT approach, the theoretical and experimental results show that the former is faster than the latter with respect to the computing speed.In a word, the proposed NAMC and the NAMTR approaches for the image representation and manipulation, as envisaged in this paper, show a very strong promise and have good potential in business applications dealing with image processing, such as reducing storage room, increasing transmission speed, and improving pattern match efficiency.