Image Classification Based On Fractal Dimension And Co-occurrence

A study shows that humankind to receive all the information that more than 80 percent through the visual that. And voice, text messaging compared to the image contains more information, more intuitive, more accurately, and so they have a higher and more efficient use of a wide range of adaptability. At present, image processing technology has been a computer science, information science, physics, biology, medicine, social sciences and even in areas such as between the various branches of study and research of the target. Image processing applications more widely, has infiltrated into the industrial, aerospace, military, medical, security and other areas in the national economy and people's livelihood, and play an important role. In the image processing technology in the image texture features analysis is a branch of one of its important image characteristics including shape, color, texture, texture of the image is a basic and important features. Texture is usually defined as a partial image of the nature of the local area or the relationship between pixels in a metric. Texture analysis is the identification image understanding, analysis and recognition of the important part of the study. Not only does it have a wide range of attention and research, in practice has also been a large number of applications. Such as in meteorology, through access to the radar images on the texture analysis to determine cloud types of meteorological forecasting; in medicine, medical imaging equipment through access to the picture of texture analysis for the medical staff to assist in economic production, The wood processing industry, through the wood surface texture analysis, we can distinguish between the different species of wood and through classification processing, wood products to achieve the best sensory effects and economic benefits.This paper estimates the early, and the recent multiple texture analysis methods, introduced Gray co-occurrence matrix and the fractal dimension of the relevant theoretical knowledge. Mainly based on the study and discussion based on the fractal dimension and texture gray symbiotic characteristics of the matrix, achieving image classification. Main contents include the following:(1) systematic analysis of fractal dimension in the roughness and texture, texture direction of the relationship between the fractal dimension that can better reflect the complexity of the image texture and roughness, and rotation invariant, etc. characteristics. Choose a paper fractal dimension of the difference in the box dimension of image classification.(2) the co-occurrence matrix gray characteristic parameters were selected pixel pitch d = 1,2,3,4,5, direction angle = 0°, 45°, 90°, 135°four directions, experiment with their own picture and the characteristics of the experimental data, select the optimal spacing and orientation angles.(3) Use Matlab and Delphi establishment of image classification system, select Stone picture as an example, in the three different types of stone images, using fractal dimension Gray co-occurrence matrix and the method of combining automatic discrimination.