The Design Of Skin Aging Evaluation System Based On Computer Vision

Skin aging is a complex process due to intrinsic and extrinsic factors. The intrinsic factors are the skin undergoes chronological changes with age. In addition, the skin undergoes environmental damage as a direct contact with the environment. For example, the photoaging from cumulative sun exposure. Skin aging is the most intuitive and obvious sign of the human aging process. Qualitative and quantitative determination of skin aging is of particular importance for the evaluation of human aging and anti-aging treatment effects.The methods for evaluating skin aging can be divided into subjective assessment through human visual or tactile sensations, complicated mechanical measurement, and simple and objective assessment based on computer vision which has gained particular attention. This paper was conducted to explore an automated system for evaluating skin aging grade with the quick and direct acquirement of skin samples by digital imaging device, the analysis and feature extraction of skin images by image processing technology, and automated evaluation of skin aging grade by pattern recognition. The detailed steps were as follows:First, the ventral forearm skin images were obtained by a portable digital microscope. Then the skin images were analyzed by image preprocessing, image segmentation and feature extraction to explore the skin texture information. Last, the values of texture parameters were taken as inputs of Self-Organizing Map(SOM) network to train the network for data clustering, and then for the automated evaluation of skin aging grade.Skin images of 120 volunteers were analyzed in this study eventually. In addition, two texture parameters based on skin furrows, i.e., mean width of skin furrows and the number of intersections formed by the skin furrows, were extracted. The experiment results showed that the segmentation accuracy of the skin furrows was high and the extracted parameters appeared to be two valid parameters for characterizing skin texture changes. Therefore, the skin texture parameters were taken as inputs of the SOM network, and the skin data were divided into 6 groups by the network, i.e., six kinds of skin aging grade were achieved. Once the training of the network done, the skin aging grade of new subjects can be evaluated by the network.The designed system of skin aging evaluation seems to be objective and quick, which can be used for quantitative analysis of skin images, and automated evaluation of skin aging grade. When there were enough skin samples, the trained network may contribute to the diagnosis and analysis of skin aging.