Automatic Registration of Aerophotos Based on Feature Space
Changqing Yang · 2005
To solve the general outliers and noise corruption problem in image registration, a new fast algorithm based on feature space is proposed. Corners are extracted using improved Harris operator, and feature space of Corner is built based on average square gradient matrix. Then three steps of matching strategy are adopted to realize image matching. The experiments show that this algorithm has the good performance under varies of conditions such as rotation, scale, and noise, especially in the condition that images overlap with less than 1/4.