Although CMOS active pixel sensor (APS) cameras are used as receivers for low-bandwidth visible light communications (VLC), they have typically not been used to receive high-bandwidth VLC signals. However, by selectively scanning pixels on the image sensor, the sampling rate of CMOS APS pixels can be greatly improved, enabling them to capture high-bandwidth signals while maintaining the ability to scale to high image resolutions. To analyze CMOS APS image sensors for potential use in high data-rate applications, the CMOS active pixel is modeled as a linear shift-invariant (LSI) system. Using this LSI model, the frequency response (as a function of device and timing parameters) and the optimal exposure time for each sample are derived. The LSI model is also used to investigate the effects of clock-jitter on CMOS APS VLC receivers.