Abstract
Based on the emerging and significant Soft DSP technique, this paper develops Soft DSP of face recognition to boost Face IDentification (FaceID) embedded system without extra coprocessor cost and power consumption. This paper not only illustrates the HW/SW codesign methodology of Soft DSP of face recognition in detail, but also gives a step-by-step implementation tutorial on Nios II embedded platform. The measurement results of Soft DSP of face recognition can verify Soft DSP technique easily achieves faster processing performance than the pure SW technique. Soft DSP technique has been entitled to be the development mainstream of FaceID embedded system field.