Based On Independent Component Analysis Method Of Image Denoising And Edge Detection Algorithm

Edges, which contain important information, are the most basic features of the image. The most important features which image segmentation rely on and the important information source of texture features are both edges information. Their advantages are able to outline the shape of the region and to be defined locally and so on. Thus the image edge detection is an important aspect of image processing, which has been one of classical research field of computer vision and image processing. However, due to a variety of reasons makes images obtained with the noise pollution, so the first step in edge detection image denoising is usually, therefore, in many areas of image denoising have occupied a very important position.With the rapid development of the national economy and improving the quality of social life, the power capacity of the industry's increasing need for stability and reliability of power supply is also getting higher and higher requirements. In order to avoid faults, it is necessary for the insulation condition of electrical equipment to carry out the necessary monitoring, maintenance and repair. Power equipment (power transmission equipment) a sense of infrared temperature measurement is a commonly used method in insulation of electrical equipment to monitor the status. Thus, the preprocessing before faults diagnosis is very important. The preprocessing generally requires image edge detection to diagnose faults, image denoising is the first step of edge detection, thus, study image denoising and edge detection have a good practice.In this paper, the traditional image denoising and edge detection method are studied, and then a new edge detection approach is introduced by summing up the deficiency of the traditional algorithms and studying the characteristics of the methods of ICA. A better result is achieved when it is applied in practical applications.The primary study of this paper can be summarized as follows:The deficiencies of the traditional image denoising and edge detection methods are summarized by studying these traditional methods, and then ICA approachs are introduced to apply in image denoising and edge detection. Several commonly used optimization algorithms are simulated for comparing tests, in order to select the best optimization algorithm to prepare.On the ground of the ICA blind signal separation(BSS)-based image denoising approach, the sparisity of base function and sparse code shrinkage(SCS) which applied in image denoising are studied, and man-made images and natural images denoising simulation results are provided respectively. By comparing with the traditional methods, signal-to-noise(SNR) shows that sparse code shrinkage denoising method in practical applications have better results.On the study basis of the sparse of ICA, the image edge detection is achieved by taking advantage of the principle of SCS based on the direction of basis function; Chapter four also provides comparison between the proposed edge detector and traditional methods, such as Sobel,Robert,Prewitt and Canny, by simulation tests and their performance is discussed as well. In addition, the non-negative sparse coding(NNSC) are introduced. NNSC is applied to detect edges, which it is proposed as a new edge detector, and its pratical performan is tested by comparing with the standard ICA edge detection method. The result shows that the proposed NNSC edge detector represents more edge informations.In this paper, the theory and application of ICA are researched deeply, and the proposed methods and its application in image processing have a certain innovation. Experiments show that these methods's with a certain degree of reference value and practical significance performance are superior to traditional methods.