Knee Infrared Image Feature Extraction Research Based On Entropy Algorithm

With the infrared thermal imaging technology more and more mature, it is widely sued in medical field. The processing and analysis method of medical infrared image are widely concerned. In this research, Entropy algorithm was proposed to process the knee joint infrared image, finally find the relationship between the knee infrared image and the information entropy value.Firstly, the infrared imaging technology and its application in medical field are analyzed, and combining the infrared spectrum characteristics of human tissue, the research foundation of knee joint infrared imaging is established. Through further research, the thermal infrared imager is used for collecting knee infrared image, and the experiments were detailed design. Finally,through comparing the characteristics of approximate entropy algorithm and sample entropy algorithm in analysis of infrared image of knee joint, sample entropy algorithm is more superior in the knee joint infrared image analysis.Basis on the analysis of sample entropy theory in this paper, the realization of sample entropy algorithm is given out. In this research, through converting the knee infrared image to gray-scale pixels series sequence, sample entropy algorithm is used to analyze the pixel series sequence. Through using the sample entropy algorithm to analyze the healthy person's knee and Patients knee, the Quantitative analysis of whether the knee joint is disease or not is expected. And it can provide basis for clinical diagnosis.Two mainly aspects are discussed in the human knee infrared image analysis problem.One aspect is the acquisition experiment of knee infrared image. Through analyzing the experiment environment and the instrument parameters, the experiment is detailed designed to obtain the appropriate infrared image for the research.The other one is the method of using sample entropy algorithm to analyze converting the knee infrared image, which is converted to gray-scale pixels series sequence. Through quantitative analysis of sample entropy algorithm, whether the knee joint is disease or not, and its lesions level is figured out.Through the analysis of the knee infrared image, approximate entropy algorithm is not suitable for this study. However sample entropy algorithm is very appropriate. The average sample entropy value of healthy persons is given out, which called reference value. When whose knee infrared image entropy value small than the reference value, its knee is possibly diseased. And the value smaller, the diseases is more serious.