Robust Automatic Multi-Camera Viewpoint Stabilization using Harris Laplace corner detection and Spanning Tree

In this paper, an algorithm to improve image stabilization process in 3D for multiple camera viewpoints from different angle is developed. It can automatically simplify the region of interest with unique key-points matching. Harris Laplace corner detection has been used to find accurate matching key points for different photometric changes, scaling in images. Spanning tree has been used to improve the connectivity of correct matching pairs by minimizing the global error value. Spanning tree is the key for stabilized randomly positioned camera viewpoints. It always gives successful stabilization output. This process doesn't need to analyze the key-points twice. The results show that the proposed algorithm can process more than 120 camera viewpoints in less than two (<;2) seconds. It shows a great consistency for different image sets for different viewpoints of cameras.