Due to the traditional hair extraction methods being easily affected by the structured noide in skin lesion regions, the texture spectrum iterative tensor voting algorithm is proposed, which combines the texture spectrum of the image with a defined attenuation functionbased on the angle deviation between the voter and receiver. It allows the proposed method to continuously update the voting scale of the voting field during the iteration process, and thereby obtains more and more voting information of the hair structure in the image. Some important submodels are introduced to enalbe the method to faster diffuse inthe early stage of iteration, and slowly diffuses in the later stages to simultaneously improve the speed and quality of the image restoration. The proposed method can optimze the running time in Chap. 3.