黑色素與紅色素影像檢測系統之設計與實作

Abstract

近年來醫美的技術不斷的在進步，同時醫美市場也在不斷的擴大，而其中有很大比例是與美白治療有關，這方面的使用人數不斷的在逐年攀升，因此如何能夠有效的鑑定美白產品的使用成效便非常重要，而觀察美白產品成效的方式不外乎有兩種：透過人體眼睛的主觀觀察與透過醫療電子器材的客觀觀察。 主觀觀察結果非常的因人而異，變動範圍大，主要在於個體間沒有一個穩定的衡量標準，因此容易使得透過人體眼睛觀察到的結果受到當下的環境、情緒起伏影響；因為一台要價數十萬的醫療分析器材非一般大眾所能負擔，因此當一般民眾想要得到客觀的黑色素數據時，則必須去美白診所諮詢，這樣是非常不便的。 基於以上原因，本論文利用影像分析的方法去近似市面上黑色素分析儀器的光學量測方法，基於影像與光學的相關性，利用皮膚組織對不同波長光的吸收及散射特性以及影像不同色彩頻道來評估使用者黑色素濃度，並在之後進行皮膚與其他背景的分辨，使得對於含有非皮膚區域照片的分析更準確。 此平台相較於其他市面上儀器具極佳的可攜性、價格低廉且可擴充性的優點，在量測黑色素濃度的同時，還會記錄使用者的量測紀錄，提供一個方便觀察膚色變化趨勢的客觀途徑。

With the advance in cosmetic medical technology as well as the expansion of cosmetic medical market in recent years, more and more people get cosmetic medical treatments, especially skin whitening treatments. For the growing number in cosmetic medical treatments related usages users, it’s important to find a way to evaluate the effectiveness of different cosmetics or cosmetic treatments. Generally, we can classify the methods of evaluation into two types: subjective observations through human being vision and objective evaluation by electronic medical devices. Results of subjective observations vary depend on individuals in a wide range. The major reason for this consequence is that there is no stable reference point for the feeling of human being. People are easily affected by the environment and state of mind at that moment. However, to acquire the value of melanin concentration objectively, people need to go to cosmetic medical clinics for consultation, which is pretty inconvenient, because they cannot afford those electronic medical devices which are cost several hun-dred thousand dollars. For reasons above, this thesis applies an image-based analysis method on smart phone platforms to approximate the results of some other skin evaluation products based on optical analysis methods. Because of the similarities between images and optics, we use the properties of different absorptions and diversions of skin tissues on different length of light and RGB color channels in digital images to evaluate the concentration of melanin in human skin. Background area and scattered region removal is conducted afterward to improve the accuracy of index value on images containing non-skin objects. In comparison with other products on market, this application on Android platform has much better portability, extensibility and much lower price. My application can rec-ord the results of analysis after each measurement and provide the users an subjective way to observe the trend of changes in skin color.