Fast global motion estimation using iterative least-square technique

Global motion estimation is an important task in a variety of video processing applications, such as coding, segmentation, classification/indexing, and mosaicing. The main difficulty in global motion parameter estimation resides in the disturbances due to the independently moving objects. The iterative least-square estimation (ILSE) technique [G. B. Rath and A. Makur, 1999] is commonly used in estimating a four-parameter model of global motion. In this paper, a modified ILSE (MILSE) technique is developed, which is capable of estimating the parameters with any number of macroblocks without considering them in order of rows and columns. The performance of the MILSE algorithm is analyzed and quantitatively and qualitatively compared with the ILSE technique. Experimental results show that the proposed technique is not only computationally fast but also robust to the disturbance caused by independently moving objects.