A combined non-uniformity and bad pixel correction method for superpixelated infrared imagery

With the recent introduction of infrared cameras that have the ability to produce variable acuity imagery, it has become necessary to develop methods for bad pixel replacement and non-uniformity correction within superpixels. Since a superpixel is formed by averaging a group of smaller pixels on chip prior to readout, producing a single value, we cannot apply gains and offsets to the individual pixels that contribute to the superpixel value, nor can we replace bad pixels within a superpixel before they corrupt the aggregate intensity of the superpixel. Without new superpixel correction methods, the imagery produced by this exciting technology is less appealing to human observers and corrupted superpixel intensities lead to problems with the algorithms that process the imagery to perform useful automated tasks, such as "hot-spot" tracking. This paper will introduce a method for performing the non-uniformity and bad pixel corrections in superpixels and demonstrate the performance.