This paper introduces a block-based motion estimation algorithm based on projection with adaptive window size selection. By using one-dimensional (1-D) projection the original two-dimensional (2-D) block-matching problem is translated to a simpler 1D matching, which eliminates majority of potential pixel participation. In this paper, we propose an efficient fast motion estimation method by using region similarity detection (RSD) algorithm to reduce the encoding time. The proposed RSD algorithm effectively estimates the texture direction of the block to find the direction and then horizontal, vertical or diagonal projection is employed for SAD calculation. Experimental results show that the proposed algorithm achieves more than a 60% time reduction with negligible PSNR loss and bitrate increments.