SPEEDING UP SIFT, PCA-SIFT AND SURF USING IMAGE PYRAMID

Abstract

Scale Invariant Feature Transform (SIFT), Principal Component Analysis (PCA)–SIFT and Speeded Up Robust Features (SURF) are common robust feature detection methods used in photogrammetry and computer vision applications. The performance of these methods have been widely investigated and compared. In terms of processing time, results show that SURF is relatively the fastest due to utilizing integral image. However, these techniques are still slow and need to be improved for nearly real time applications, such as those based on vision navigation. This paper works on speeding up SIFT, PCA-SIFT and SURF using image pyramid. The images are firstly resampled and matched to detect the interest points. Then, the approximate locations of the matched points are determined on the original images from similar triangles. These points are surrounded by small searching windows and matched again with the corresponding searching windows in the other image. As a result, instead of matching the whole two images, a number of tiny images are matched together. The results show that the idea is powerful for reducing the processing time of such techniques significantly. The performance of this idea is affected by the resampling level and method, the image size, and the selected number of matching points.