Research On The Quantitative Analysis System Of Skin Lesion Detection Based On Vision

Skin is an integral part of the human body and is closely related to health.In the current situation of high incidence of skin diseases,traditional skin lesions detection methods are time-consuming,laborious and subjective,and it is difficult to establish a unified index.Therefore,it is of great significance to develop a convenient,reliable and objective rapid detection system of skin lesions,which can reduce the burden of practitioners and carry out long-term monitoring of skin diseases.This paper takes white spot lesions as the research object,and develops a quantitative analysis system of skin lesions detection based on vision.Firstly,the skin lesion image was collected by ordinary camera,and the main factors affecting the skin image quality were studied,including hardware error,illumination and noise,and the corresponding methods to reduce the interference were put forward.Secondly,the characteristics of four algorithms,such as Otsu method were analyzed by experiments,and which of them were changed to improve the accuracy of the algorithms for segmenting white spot image.Then,in order to obtain more stable and robust segmentation results,the segmented subsets of different algorithms were synthesized in Markov random field,and the pixel points were redistributed according to the weight factor,so the combined segmentation result was obtained.On this basis,the area and color features that are used to evaluate the white spot lesions were extracted quantitatively.Finally,the system design was implemented on the LabVIEW platform.The experimental results show,the skin lesion detection system designed according to the research results of various parts,that can be used for dealing with the image of the white spot skin lesion,such as the multi-algorithm segmentation,subset fusion and the feature extraction,so the rapid measurement of the skin lesion information is realized.