Noise analysis in real-time portal imaging. I. Quantization noise.

The limit at which quantization noise becomes dominant in video-based real-time portal imaging has been studied. Quantization noise due to truncation in integer frame averaging is shown to be dominant over the input analog-to-digital converter (A/D) quantization noise, unless image addition is used in video-based real-time portal imaging systems. Portal images acquired with the Newvicon camera by averaging more than 64 frames are found to be dominated by the quantization noise due to truncation. It has shown that the signal-to-noise ratio (SNR) is limited to 886.8 when using an 8-bit A/D with digital frame averaging, but higher values can be achieved with digital frame addition. It is also shown that digital frame addition together with 16-bit processing can achieve higher contrast resolution than digital frame averaging and 8-bit processing.