| In recent years, Complementary Metal-Oxide-Silicon (CMOS) image sensors have made rapid advances and gained favor over charge coupled device (CCD) image sensors in many fields due to low cost, higher levels of integration, lower power consumption and high programmability. Typical image sensors, however, do not have sufficient dynamic range (around 60dB) to capture the full range of illumination in the real world (over 120dB). In this dissertation, three different types of wide dynamic range CMOS image sensors are introduced.; The first sensor proposed in this dissertation increases dynamic range (DR) using the self-reset with pixel level analog-to-digital converter. The proposed self-reset technique increases the well capacity of the image sensor. Consequently, the proposed sensor can increase not only DR but also signal-to-noise ratio (SNR) in contrast to the other wide dynamic range CMOS imagers.; The second one is a dual-mode wide dynamic range CMOS image sensor (CIS) that has two different operation modes: logarithmic and floating-point mode. Each operation mode has its own advantages and disadvantages. The proposed dual-mode sensor can choose the operating mode manually or adaptively for the best performance.; The last one is a floating point CIS. A detailed analysis of the proposed floating point CIS is provided. Based on the analysis, an exponent error tolerant method is proposed. Assigning a small portion of the well capacity as the safety zone can effectively remove the exponent detection error, which will increase the SNR of the CIS. |