Motion estimation based on triads of Gabor filters: DSP board implementation

The quality and rapidity of motion estimation in image sequences are fundamental in many applications as for instance in artificial vision and three-dimensional scene reconstruction. Among all the existing techniques, we are particularly interested in energy-based methods which are known to furnish high-quality results but usually require intensive calculation. We present a fast energy-based method which combines in a direct manner the energetic responses of Gabor spatio-temporal filters organized in triads and we describe an implementation of this technique on a general purpose DSP board. Our hardware implementation attains a reasonably fast output rate (few images/second). These results open interesting perspectives for real-time implementations, such as in mobile robotics.