Research On Texture Feature Of Medical Image Based On MPEG-7

There are a large number of digitized medical images in the computer produced in hospital, such as X-ray photograph, CT, MRI, US, SPECT and PET and so on. It can save too much image storage space and improve the efficiency of retrieve. How to effectively retrieve medical images and make amounts of medical images service for medical research and clinical diagnose, have become the urgent needs and the hot researches in the field of medical image. The existing medical image system can only provide the patient name and serial number for retrieval methods, which can not fulfill the doctor's clinical needs. For example, when doctors encountered puzzling cases, we need retrieve some medical pictures according to a number of similar medical cases for the clinical diagnose. MPEG-7 was regulated for meeting the needs of internet by ISO/IEC. It is used to fix format to describe the characteristics of the image or sound data. For fast retrieving results, the image maker can save MPEG-7 characteristics of information to image.This article uses the Homogeneous Texture Descriptor and Texture Browsing Descriptors to provide interfaces to retrieve the image, to meet the versatility and reusability. Homogeneous Texture Descriptor and Texture Browsing Descriptor are two of MPEG-7 texture descriptors. They can retrieve texture descriptors from deference way. The HTD can describe image from texture thickness and direction. The TBD describe image from thickness, direction and regulation. We use the weight to make they adapt more kinds of medical image.Through the experiment in medical image retrieval platform, the retrieval results of two algorithms are better than the classical texture algorithms- gray-scale matrix and generalized gray-scale matrix. They own availability and versatility. But its need some elevation in the retrieval accuracy and retrieval time in the next step.