Design of Scalable Hardware-Efficient Compressive Sensing Image Sensors

This paper presents a new compressive sensing (CS) measurement method for image sensors, which limits pixel summation within neighbor pixels and follows regular summation patterns. Simulations with a large set of benchmark images show that the proposed method leads to improved image quality. Circuit implementation for the proposed CS measurement method is presented with the use of current mode pixel cells; and the resultant CS image sensor circuit is significantly simpler than existing designs. With compression rates of 4 and 8, the developed CS image sensors can achieve 34.2- and 29.6-dB PSNR values with energy consumption of 1.4 and 0.73 μJ per frame, respectively.