Dynamic range enhancement algorithms for CMOS sensors with non-destructive readout

Image sensors capable of a non-destructive readout (NDR) allow reading several frames during the integration time without affecting a photo charge being collected in the pixelpsilas well. We show that two published algorithms for processing the sequence of NDR frames into a high dynamic range (HDR) image suffer from ldquostepped gradientrdquo artifacts using data taken with an ldquooff-the-shelfrdquo CMOS image sensor. We present a novel algorithm that is capable of alleviating this image artifact that is due to suboptimal saturated pixel treatment. Our procedure works for both black and white and color images. In addition, we present joint demosaicing and denoising for color HDR image reconstruction and its performance is compared with conventional demosaicing algorithms. The overall performance of the combined methods provides a DR increase of up to 44 dB compared to raw data. We demonstrate our algorithms using images taken with a black-and-white and color NDR capable camera system.