Research On White Hair Detection Based On Image Processing Technology

Personal image is very important for individual life, a good personal image willleave a good impression to others, promote the development of social activities andmake physical and mental pleasure to people. On the contrary, a undesirable personalimage will give a negative impact on life and work, usually affect people’s mental health.White hair has brought a lot of serious problems to young and middle-aged people,especially the personal appearance of women. The traditional solution is make whitehairs dyed black by using hair dye. Many experiments have proved that long-term usingof hair dye can cause human body skin allergic reaction, and it’s much easier to absorbchemicals if human have any trauma or skin disease, which may result in theaccumulation of chronic poisoning, induce cancer, pregnant women may even occurfetal abnormalities.With the development of computers and artificial intelligence, people increasinglywant to accomplish tedious repetitive work have a huge amount by intelligentmachinery. This paper presents a vision: to people who have a small amount of whitehair, extracting the roots of white hairs with the head image processing by targetextraction algorithm, be cut off or single coloring according to the proportion of whitehairs, can remove white hairs healthy and efficient. Contents of this article are based ondigital image processing techniques to detect white hair, in order to lay the foundation toachieve associated electrical and mechanical devices in the next step.This paper designed a small amount white hairs identification method based onimage target detection, including image preprocessing and target identification. Theimage preprocessing include: image illumination equalization, image noise and trashremoval, the target recognition include edge detection and boundary tracking. Theauthor used the collected head images to verify each step. Through experimentalcomparison, we take image illumination balance by homomorphic filter, and theappropriate high-frequency and low-frequency component to filter is selected. We selectthe two-dimensional haar wavelet packet for image decomposition, imagereconstruction, remove noise and impurities. This thesis analysis and comparison theprocessing effects of a variety of edge detection operator, choose an improved Sobeloperator for its better detection results, merge and extract the target for the edge byautomatic machine based on boundaries calibration algorithm which has a lower timeand space complexity. This thesis achieves the algorithms above by Matlabprogramming, and verified the effectiveness of the algorithm through experiments.