Improved image accuracy in Hot Pixel degraded digital cameras

In our previous papers we concluded that the main source of defects in digital cameras are “Hot Pixels” and showed that their numbers increase at a nearly constant temporal rate (during the camera´s lifetime) and they are randomly distributed spatially (over the camera sensor). The defect characteristics of each hot pixel, i.e., the offset and slope of its dark response, appear to remain constant after formation and can, therefore, be extracted and used for image correction. In this paper we suggest a novel method for correcting the damage to the image caused by a hot pixel, based on estimating its dark response parameters. Based on experiments on a camera with 31 known hot pixels, we compare our correction algorithm to the conventional correction done by interpolating the four defective pixel neighbors. We claim that the correction method used should depend on the severity of the hot pixel, on the exposure time, on the ISO, and on the variability of the pixel´s neighbors in the specific image we are correcting.