In this paper, we propose a fast Motion Estimation (ME) algorithm for low computational complexity based on spatial correlations. The proposed algorithm uses the motion vector of neighboring blocks, to decide the location of initial searching point. The number of searching point decreases by changing the location of initial searching point. Experimental results show that the proposed algorithm reduces searching points about 7 ∼13% and increases PSNR about 0.1∼0.5(㏈) compared with conventional algorithm.