| Most of today's video and digital cameras use CCD image sensors, where the electric charge collected by the photodetector array during exposure time is serially shifted out of the sensor chip resulting in slow readout speed and high power consumption. Recently developed CMOS image sensors, by comparison, are read out non-destructively and in a manner similar to a digital memory and can thus be operated at very high frame rates. A CMOS image sensor can also be integrated with other camera functions on the same chip ultimately leading to a single-chip digital camera with very compact size, low power consumption and additional functionality. CMOS image sensors, however, generally suffer from lower dynamic range than CCDs due to their high read noise and non-uniformity. Moreover, as sensor design follows CMOS technology scaling, well capacity will continue to decrease, eventually resulting in unacceptably low SNR.; In this dissertation, new pixel architectures and algorithms are presented that enhance the dynamic range and SNR of CMOS image sensors by utilizing their high speed readout and integration advantages. The thesis is divided into three parts. First, a 352 x 288 Digital Pixel Sensor (DPS) chip with per-pixel single-slope ADC and dynamic memory fabricated in a standard digital 0.18um CMOS process is presented that demonstrates the high speed potential and scaling advantage of CMOS image sensors. The chip performs “snap-shot” image acquisition at continuous rate of 10,000 frames/s or 1 Gpixels/s. Second, an algorithm based on statistical signal processing techniques is presented that synthesizes a high dynamic range, motion blur free, still image or video sequence from multiple image captures. The algorithm is recursive and consists of two main procedures—photocurrent estimation and motion/saturation detection. Photocurrent estimation is used to reduce read noise and thus to enhance dynamic range at the low illumination end. Saturation detection is used to enhance dynamic range at the high illumination end, while motion blur detection ensures that the estimation is not corrupted by motion. Motion blur detection also makes it possible to extend exposure time and to capture more images, which can be used to further enhance dynamic range at the low illumination end. The algorithm operates completely locally and recursively, its modest computation and storage requirements make the algorithm well suited for single chip digital camera implementation. Finally, to solve the problem with CMOS technology scaling and further enhance sensor SNR at high illumination, a self-resetting scheme is presented. In this scheme, each pixel resets itself one or more times during exposure time as a function of its illumination level, resulting in higher effective well capacity and thus higher SNR. The photocurrent estimation algorithm is then extended to take new noise components into consideration, and simulations results demonstrate significant dynamic range and SNR improvements. |
