【作者】 陈峰；

【导师】 荆涛；

【作者基本信息】 北京交通大学 ， 通信与信息系统， 2008， 硕士

【摘要】 本文对人脸皮肤表面参数进行了分析探讨,并建立了以数据管理和分析为一体的研究平台:数码智能皮肤分析系统,该系统分为数据采集、数据分析、数据管理和用户界面三大主要的模块,通过对人脸皮肤图片分析,得出肤色、斑点、水分油份、纹理皱纹和毛孔几大指标,对皮肤表面状况进行了综合的评价,并完成了该系统第一阶段的迭代开发。本系统采用带USB接口的采集仪器进行图像采集并将其组件化,建立了基于ADO访问技术的数据管理功能,利用封装于DLL的识别算法对皮肤图像进行快速综合分析,同时采用友好、图形化的分析界面,使使用者能够对皮肤当前状态有直观的全面的认识,为进一步提出客观、重复性好的皮肤分析打下坚实的基础。文章也对系统中需要解决的图像处理问题做了初步的研究。在肤色检测中将理论和实际研究对象相结合改进了分类的阈值,在皮肤图像纹理的提取中,对传统的方法进行了改进和优化,在水分油份检测中,利用了统计回归方法建立相关模型取得了良好的效果。更多还原

【Abstract】 This paper analysis and discuss skin surface parameters, then establishes a data management and analysis platform for the integration of research: digital intelligent skin analysis system, this system is divided into data acquisition, data analysis, data management and user interface three main modules, through analysis images of facial skin, come to color, spot, oil, moisture, wrinkles, texture and pore several major indicators of the state of the skin surface comprehensive evaluation, and completed the first phase of the system Iterative development.This system uses a USB interface with the collection apparatus for image acquisition and its components, the establishment of technology-based ADO visit to the data management function, the DLL using packaging in the skin image recognition algorithm for rapid analysis, using friendly, graphical analysis interface that let users have intuitive comprehensive understanding for the current state of the skin and for the further objective, reproducible analysis of skin and lay a solid foundation.The article also done a preliminary study to those problems that need to be addressed in the system of image processing. In color detection, with combination of theoretical and practical object ,we improved the classification threshold, In detection of the face image texture, we improved and optimized the traditional methods, and used statistical regression Methods correlation model has produced good results in the oil and moisture detection.更多还原