Research On The Matching Technique In 3D Reconstruction

Matches between two or multiple images that belong to the same scene are the foundation problem in computer vision, and also the key step in the process of the 3D reconstruction. The matching efficiency and performance deficiencies in 3D reconstruction are obvious, so research on this kind of matching method is a focus in the field of image, especially the research on feature-based matching method is of great practical significance.One difficulty of the feature-based matching method is feature extraction technique. In 3D scene reconstruction, whether the feature extracted from simple image or complex image, how to find a good feature extraction algorithm to extract specific image feature is urgent need. In the research of extracting a simple image feature, SUSAN operator and Harris operator are widely used. The advantages and disadvantages of these algorithms and some problems that needed to pay attention to in application are raised by detailed analysis and experiment test. In the research of extracting a complex image feature, in view of the matching efficiency and performance deficiencies of the SIFT keypoints descriptors, firstly more keypoints information which is used for feature description is acquired, and then keypoints descriptor is compactly constructed with dimensionality reduction technique, which makes the final structure of the descriptors is both distinctive and robust. The advantage of such improved algorithm is that it gains much more keypoints information and takes the matching efficiency controlled by dimensionality of descriptor into account at the same time, which results a higher matching performance and a better matching efficiency. Besides, more bad matches appear in the complex 3D reconstruction due to noise, block and the negative threshold selection. Epipolar geometry method is used solve that problem. Firstly matches among images are extracted by IPC-SIFT, and then RANSAC robust method, non-linear estimation and guided matching are used to eliminate these bad matches and good matching results are available.The above method provides a wholly new method to effectively implement the matching technique in 3D reconstruction. The advantages of them are on the wide use in many practical areas and obviously improved matching efficiency and performance, which is very important for many applications.