THE HUMAN SKIN EVALUATION AND VISUAL ASSESEMENT WAY APPLIED BY SPECTRAL IMAGING AND LAPLACIAN FILTER PROCESSING

摘要

The human skin is one of the difficult objects to get high accuracy measuring resulted by ordinary contact type spectrophotometer. As the reason, the human skin surface has extremely complex structure and half-transparency phenomenon. And each wavelength image is different. In this study, the human skin measuring way was applied as non-contact type spectral imaging, and the evaluation way was applied as Laplacian filter image prospecting. A gonio-photometric spectral imaging system was applied to measure spectral reflectance and texture of human skin. It was composed of white LED illuminates, a liquid crystalline tuneable filter (LCTF), and CCD imaging device with Peltier cooling unit. Illuminating direction was 20 degrees from normal direction, and detecting direction was normally against to samples, and the CCD device captured the images via LCTF. Measuring area was around 10cm by 7.5cm and pixel resolution was 772 by 580 pixels in this area, and resolution was 192 dpi. To get high accurate gonio-photometric reflectance spectrum and imaging information, each wavelength sample image was compensated by measuring of black/white lattice pattern printed plate to sense small shift amount of x and y direction before measuring human skin. And laser projector was applied to get 3-Dimension human skin object by lattice pattern projection to compensate optics geometry. Measured part of human skin was in the middle of lower arm and measured 1 female subject. This subjects' age was twenty. After measuring, Laplacian filter value of each 10nm wavelength from 420 to 700nm for measured spectral imaging data was calculated with 1 by 1 to 50 by 50 pixel size, while moving average to analyse correlation between spatial frequency and subject. Also tristimulus and L*a*b* value were calculated from each pixel spectral imaging data under illuminant D65, F10, A, and three types of LED. As the result, Laplacian filter value, spatial frequency, and wave-length was shown in this study. This colorimetric way was quite useful and highly possible for evaluating human skin characteristics. To research on the correlation between visual image evaluation and Laplacian filter value, the human visualizing assessment test was applied as the SD method and analysis of variance. From the results of appearance assessment, we figured out that the apparent order of light sources which gives comparatively higher score to positive appearance emphasis from LED A, A and F10, and the other order of light sources which gives comparatively lower score starting from LED C, D65 and LED B exist. Especially, LED B showed very low relationship.