Image Mosaic based on Image Entropy and the Number of Feature Points Control

摘要

In order to reduce the time of the process of feature description in SURF algorithm and improve the speed of SURF algorithm, an improved SURF algorithm based on image entropy is proposed. It uses image blocks in the overlap region of original images. Then the image entropy of each block is calculated to select reliable area. The reliable area is only used to detect feature points. The two methods (Uniform sampling and the distance constraint between feature points) are used to control the feature points number. The speed of SURF algorithm is increased by about 65% and the performance is also improved.