Temporal and spatial noise correction oriented to the maximization of contrast and compensation of ilumination in infrared images

Abstract

he infrared images have defects associated with the hardware such as defective pixels, temporal noise, spatial noise, vignetting and so on. They are inherent to the environment as illumination changes, currents of convection and conduction. This work makes use of methodologies focused on the correction of non uniformities as it is the algorithm of constant statistics (CS) and constant rank (CR), likewise  using mathematical morphology and mapeos of contrast (MC) with the aim of maximizing the contrast and compensate the illumination changes from each of the infrared frames.

The metrics used to evaluate the performance of the techniques correspond to RMSE (half quadratic error) and the index of merit Rho which is an approximation of the amount of noise removed, the results obtained with the use of the correction techniques of non uniformities have an index Rho and a low RMSE compared to the values of these metrics for contaminated frames, Being these frames close to the metrics values obtained for a frame without spatial noise,  we can even obtain better results when we apply morphological operators and mapeos of contrast to the sequences of images that come out of the stage noise correction given that the image contrast is improved adaptively.