A novel grating matching method for 3D reconstruction

摘要

Automatic 3D surface reconstruction has been an important research topic in digital photogrammetry for many years. Shaping from stereoscopic information is one of the widely studied topics in computer vision Its central part is to solve the stereo matching problem automatically. Most algorithms used to solve the matching problem can be categorized as either area-base techniques or feature-based techniques. Feature-based techniques have gained more and more popularity for and it is the method that supports activities in object recognition and image understanding. But none of the former research can ensure the 100 percent matching exactness, thus, cannot complete high precision 3D surface measurement. In order to improve the precision of 3D measurement, we design a novel grating matching method, which can ensure 100 percent matching accuracy. It does not need any other assistant symbol or flag, only select one of the gratings which have to be projected to the objects. On the beginning of the measurement, the background of the object is captured by a CCD camera. Later gratings include single grating and group gratings will be projected to the object in sequence. All the images include gratings will make a subtraction with the background. The difference of the two images will be treated as 3D cues to acquire the 3D shape. Because no recognizing work is needed, the location of the single grating and each one in the group can be positioned exactly. From a lot of experiments, the proposed grating matching method is proved and it is a technique with high precision, low costs, easy operation, and an automatically matching method. Furthermore, it can be widely used in most of 3D vision recovery systems.