Wide Dynamic Range

Loosely defined, a camera's dynamic range is the difference between the maximum and minumum acceptable signal levels.  If part of a scene is illuminated too low, chances are there are not enough photons coming from that area to be converted to a meaningful electronic signal.  The detail in the dark won't  be "seen" by the camera.  Conversely, if part of a scene is very strongly illuminated (e.g. sunlight streaming through a window), the image from that area of the scene could be washed out.  In the worst-case scenario, the scene could, and often does, contain areas of extremely low and high levels of illumination.

Dynamic range is the ability of the imaging chip to convert light information.  The wider a camera's dynamic range, the better its ability to deal with these extreme lighting contrasts.  To implement XF-Dynamic feature, Bosch's 15-bit Dinion DSP uses a function known as histogram control to stretch out the details in the image by amplifying minute variations in illumination (see example).  The illumination of every pixel in a scene acquired by the CCD is recorded in a luminance historgram which sorts the pixels into 32,767 gray levels - ranging from black up to maximum scene brightness (see graph "A" below).

The camera's transfer function (the curve defining output as a function of input), instead of being a simple straight line like a normal camera, automatically increases more sharply in regions of high information density and less sharply in regions of low information density (graph "B" below).  This creates a more evenly distributed luminance histogram at the camera output (graph "C" below).  Low-contrast scenes are typically characterized by a very high pixel count at just a few gray levels.  In such scenes, the histogram reduces the number of pixels at these gray levels.  At the same time, the pixel count is increased at other (less well-represented) levels to stretch the overall picture contrast and bring out fine picture details.

WDR Diagrams

Fig. 2 - The pixels in a scene are automatically grouped by gray level into a luminance histogram (a).  This is modulated by a camera transfer function that varies with information density across the scene (b) to create a more even distribution of gray levels (c) and accentuate scene details that would otherwise be lost.

Fig. 3 - Block diagram of a 15-bit DSP CCTV camera

PREVIOUS: Digital Signal Processing Learning Center Start NEXT: Smart Exposure
Related Links
 
Top of page