Medical Image Filting Based On Image Processing And Neural Network

Digital image processing is also known as computer image processing, it means the image signals converse into digital signals and using the computer for processing.Digital image processing is easy to achieve non-linear processing, process procedures and deal with variable parameters, so it is a universal, high precision, flexible processing, information preservation, transmission reliable image processing technology. Image processing, the input are low quality images, the output is to improve the quality of after images, commonly used image processing methods have image enhancement, restoration, coding, compression and so on.One of the digital image processing main contents of the study is image restoration, it calls for must have a certain understanding of the reasons about low-quality images, general speaking,it should establish a "drop transfer model" based on the process of lowering the quality image, and then using some sort of filtering method, to restore or reconstruct the original image. From new development in recent years with the fuzzy pattern recognition and artificial neural network pattern classification in image processing are also more and more attention.Images are the main source of information acquisition and exchanging for human beings, digital image processing is widely used in biomedical engineering and very effective. In addition to CT technology, there is one type of analysis and processing for medical micro-image, such as erythrocyte and leukocyte classification, chromosome analysis, identification of cancer cells. In addition, X-ray images for the lungs increasing the clarity ultrasonic wave image processing, ECG analysis, stereotactic radiotherapy, like these medical diagnosis have wider application of image processing technology.Combined with the practical application of the necessary, the paper focus on medical image noise filting, including the following:(1) We made an in-depth analysis about the frame of kernel-based methods.The methods work as the same way.The program are adjusted to accepted inner products of the input data. Then the kernels are used to compute inner products of the data which have been mapped into the characteristic space.So the algorithm is proved to be feasible in high-dimension space.This flow manifest the modularity of kernel methods;(2) depth analysis of the existing noise filting method, although the existing noise filting methods can deal with different noise models, for medical image processing have some laws to follow. This paper focus on medical images for the impulse noise and Gaussian noise, how to carry out effective noise filting;(3) the analysis of different image noise models, this paper presents a more effective noise filting method. Experimental results show that the proposed method for medical image noise filting is relatively effective.