期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2004
卷号：XXXV Part B5
页码：311-316
出版社：Copernicus Publications
摘要：A multi-camera close range photogrammetric system for robust and precise measurement of human body surface in general and three dimensional evaluation of bedsores in particular has been designed and constructed. MEDPHOS (MEDical PHOtogrammetric System) consists of four digital synchronized cameras the optical axes of which are convergent in a pre-specified point. A light projector is fixed in the center of the cameras-carrying rig. The projector serves as a texture provider that projects a special pattern on the surface to be measured to compensate for the lack of texture in human body surface. Determination of shape, size, and density of the pattern dots was also investigated. The proposed algorithm consists of the following steps: Both cameras and projector were calibrated so that the positional and orientational parameters of the entire system are known. A new method for image segmentation and thresholding using morphologic operators was used for the detection of the projected pattern dots from the uneven background of the images. A watrershed transformation was also applied to tackle the problem of overlapping pattern dots. Then to reduce the effects of specular reflection of light due to the humidity of the object (in the case of wounds), a novel homomorphic transformation was developed and applied to the images. After segmentation of the images, connected component labeling procedure was used to set up the actual matching points. The centroids of these components were precisely calculated. Intensity-based image matching had already been tested that - due to great deviation from the Lambertian assumption necessary for solving the correspondence problem - did not give satisfactory results. So, a new algorithm based on four focal constraint has been developed that makes it possible to carry out the matching procedure without being dependent on intensities of the pixels and without a need for approximate values of the unknowns. This robust and ever valid geometric constraint is found to be very effective with non ambiguous results provided the necessary conditions for the system configuration are met. The fourth camera and the calibrated projector- that is treated like an active camera- serve as additional sources of information for consistency checking of the results. Almost all kind of information in the filed of biometrics can be obtained rapidly, robustly, and easily using MEDPHOS
