Analysis And Assessment Of Image Quality With Latent Semantic

Under the backgroud of information technology is widely used in image, image quality assessment becomes a broad and basic problem. No reference image quality assessment is the use of computers instead of human eyes in image quality assessment, and this tecknology is widely used in military, traffic and other fields. So the no reference image quality assessment has important academic value and broad application prospects.Probabilistic Latent Semantic Analysis (PLSA) originated in the classification and retrieval of documents, using statistical methods to describe the relationship between the "documents-latent semantic-words", and its essence is make documents and words are mappled to the same semantic space. In fact, image quality assessment is visual perception of person, also can be calssified as semantic analysis, and it is mean that the image can be described by visual words, also map these visual words to latent semantic space, through the distribution of the latent semantic to evaluate the quality of image.This thesis depth analysis of no reference image quality assessment based on PLSA, mainly studies the calculation method of low-level visual features for different distorted images. Respectively puts forward the adaptability wider noise extraction method and simple calculation, high efficiency of JPEG compression feature extraction method. By blur feature, noise feature and JPEG compression feature consitute a three-dimensional feature and applied to PLSA framework. It is implements that mulitple distortion type of no reference image quality objective assessment, and it obtained the good experimental results.This thesis completed another important content is that designed and programed to realize the image quality assessment system for research and development. On the basis of demand analysis of system, it designed the system of train and test function module, data collection of operation process, showing the results, etc. Using the development way that VC++combined with OpenCV to implement image quality assessment system based on PLSA. Since the core algrithm using C language programming, and it is solved the problem for parameter adjustment and operation efficiency.