基于支持向量机的图像去噪和图像质量评价的分析-analysis of image denoising and image quality evaluation based on support vector machine.docxVIP

Based on Support Vector Machines Image Denoising and Image Quality Assessment ResearchAbstractIn recent years, along with the informationization high speed development, the imagery processing as well as the application was day by day widespread, however issues arising with the attendant, such as image noise reduction and image quality evaluation is hot research topics in the field of image processing.The image in gathering, the compression, the transformation and in the transmission process, can receive own influences frequently and so on image formation equipment as well as external environment noise jamming, therefore the majority of realistic images all are noised. In order to facilitate subsequent processing and application, there is a need for noisy image denoising and quality evaluation. Good noise removal method is without prejudice to images subject to the important details, as much as possible to remove noise, and good quality evaluation algorithm is required evaluation results that are consistent with the human subjective experience. Both the image noise and the image quality assessment, we all can be understood as a process of classification. Image denoising was separated from the original image and noise; image evaluation is the evaluation of the image correctly classified to see which category it belongs to the quality level.Support vector machine (support vector machine, svm) as a typical classification, it has many advantages, can be widely used in the classification problem for dealing with small sample, high dimension and generalization performance. In view of this, this article will support vector machine method for image denoising and image quality assessment. This paper proposes a coefficient based on wavelet image denoising method and support vector machine. According to the nature of the wavelet coefficients, the paper selected as the average neighborhood of wavelet coefficients based on eigenvector for training, and then use the training recei