| Summary: | 碩士 === 南台科技大學 === 資訊工程系 === 101 === The skin condition affects people’s self-confidence, and it is also the symptom for many diseases. In this thesis, we handle the skin inspection problem. The skin data is extracted by a numbers of medical image processing techniques from the skin images in companion with the expertise from professionals. We have also developed a computerized detection and scoring algorithm to achieve the skin inspection goal.
We designed a facial image acquisition device to acquire the skin images with normal light and UV light. A series of image pre-processing steps including skin color classification, image rotation and correlation, image alignment, morphological noise removal and region of interested (ROI) detection, are performed. For skin inspection purpose, three different sizes of ROIs are extracted from the forehead region, the eyes region, the nose region, the cheek region and the chin region. These ROIs are assigned with a score by the professional experts on the basis of the skin conditions for acne, spot, wrinkle, black eye and sunburn symptoms. The features used in this study contain color and texture features. To reduce the high feature dimensions, the principal component analysis (PCA) is used. A support vector machine (SVM) training algorithm is developed to inspect the skin condition.
In this study, there are total 80 images are acquired with normal light and UV light. Each image is divided with 11 region of interested. Professionals assign each ROI with a score from 1 to 5 according to the severity of the skin diseases. For the skin image set, there are 50 images are used for training, and the other images are used for testing. The experimental result shows that the accuracy rates for the five symptoms are between 88% to 92%, and the average score differences are between 1.2 to 1.9. In this thesis, we have developed a cheap and effective skin condition inspection and scoring system to help people understand their own skin condition and provide a valuable reference for further treatment.
|
|---|
