The information contained in this document is not subject to Export Control Regulation in Canada.
Calibir GXM640 IR Camera User's Manual
Features and Configuration Options
47
Defective Pixel Correction
The Calibir GXM640 camera uses a defective pixel map to identify bad pixels in the
sensor. These bad pixels are replaced by the average value of neighboring valid pixels.
The defective pixel map is an 8-bit bitmap file (.bmp) that uses two pixel values; 0
(black) indicates a good pixel, non-zero values indicates a bad pixel.
Figure 12: Defective Pixel Map
Two defective pixel maps are available: a factory calibrated set and a user-defined map.
The factory calibrated defective pixel map identifies defective pixels across the full range
of ambient operating temperatures. The user-defined map adds pixels to the list of
defective pixels if additional pixels become defective during the camera’s life after factory
initialization.
The user-defined map can be updated manually or using automatic defective pixel
detection. Automatic pixel detection updates the user-defined map by launching a
sequence that compares each pixel value with the average value of its neighbors.
To generate valid results, a uniform scene should be presented to the camera; the
internal mechanical shutter can be used, or alternatively a uniform scene can be
presented in front of the lens (for example, a sheet of paper, or a cardboard box;
additionally, it can help to adjust the lens completely out of focus).
Configurable parameters include the size of the neighborhood window and the threshold
used to identify bad pixels (the number of standard deviations from the average value).
The Defective Pixel Detection Window Size (defectivePixelDetectionWindowSize) and
Defective Pixel Threshold features (defectivePixelDetectionResponseThreshold)
determines the algorithm behavior. In general, for very uniform images, a larger window
size provides more statistics; for less uniform images, a smaller window is recommended
(a smaller window also results in faster execution time). The threshold sets the number
of standard deviations from the normal pixel response beyond which a pixel is considered
defective.
Use the Detect Defective Pixels feature (defectivePixelDetectionTrigger) to perform
defective pixel detection.
