摘要
压缩感知是近年来发展迅速的一种新型信号处理理论,以此为基础发展的成像系统具有系统规模小、成本低等特点而受到广泛的关注。利用表征反射光强度信息的图像灰度值是物体表面属性及几何形状综合反映这一光度学基础,引入以随机采样及非线性重建为技术手段的压缩感知理论,提出了一种基于压缩感知的光度立体视觉三维成像方法。该方法通过调换探测器与光源的位置,采用单像素探测器代替传统CCD成像,从不同位置的单像素探测器采集经调制的测量数据,利用压缩感知算法重建二维目标的图像,而后利用图像中的阴影信息求解目标的表面法向量和三维表面模型。最后,利用该方法对人像和简单几何体进行三维成像实验,结果表明该成像方法能在数据欠采样的情况下,实现目标的高效三维重建;在实验基础上,进一步分析了图像数量、采样率、重构方法、探测器位置等方面对三维成像质量的影响。
Compressive sensing is a new type of signal processing theory which has developed rapidly in recent years.With the advantages of simple system structure,less data collection and super-resolution, compressive sensing imaging system has attracted a wide spread attention.Using the photometry basis that the image gray scale value indicating the reflected light intensity information is a comprehensive reflection of the surface properties and geometries of the object,this paper proposes a photometric stereo three-dimen- sional imaging method combining compressive sensing theory of random sampling and non-linear recon- struction..The method exchanges the position of detectors and light source, uses the single pixel detector instead of the traditional CCD imaging,collects the measured data of the target modulation from different positions of single pixel detector,then reconstructs the 2D image with compressive sensing algorithm and uses the shadow information of the image of single pixel camera to compute the surface normal and three- dimensional surface model.The proposed method is used to perform three-dimensional imaging experiments on the portrait and simple geometry object.Experiments show that the proposed method can achieve efficient three-dimensional reconstruction with sub-sampling data.The influence of number of directions,sampling rate, reconstruction method and detector position on the three-dimensional image quality is further analyzed.
出处
《遥感技术与应用》
CSCD
北大核心
2017年第3期449-458,共10页
Remote Sensing Technology and Application
基金
国家863计划项目"强度关联遥感成像外场综合测试技术"(2013AA122904)
中国科学院
国家外国专家局创新国际团队项目"基于强度关联的新型遥感机理与方法研究"(2013AA1229)
关键词
三维成像
光度立体视觉
压缩感知
质量评价
Three-dimensional imaging
Photometric stereo
Compressive sensing
Quality assessment
