Twelve Bit Acquisition, the Next Step in Digital Broadcast Cameras

CCD image sensors are capable of delivering at least 600% of linear video to the signal processing. This headroom is needed to reproduce large scene contrasts in a convincing way on displays not capable to handle these large contrasts. — Today's cameras have 10 bit A/D conversion in the video path. This calls for some form of nonlinear analog processing like (pre-)gamma or (pre-)knee to handle these 600% signals. This weakens the well known arguments of better accuracy and stability for these digital cameras: non-linear circuits like knee, gamma and whiteshading multipliers are not the most accurate and stable analog circuits. — The move to 12 bit A/D conversion in the camera allows digital videoprocessing with sufficient dynamic range, and all non-linear processing in the digital signal path. — This also opens the way to reconsider processing, given the new possibilities of the digital domain. Examples will be given in the paper. Some items that will be addressed are: dynamic pixel correction, automatic skin contours and a new digital highlight compression with improved color fidelity. — The architecture of a 12 bit camera, the Philips BTS LDK20P will be explained in the paper, together with the algorithms used.

Published

1996-10

Content type

Original Research

DOI

10.5594/M00112

ISBN

978-1-61482-925-6