Image processing technique using high frequency data analysis to filter low frequency data by dynamic kernel adjustment

Abstract

A method includes separating image data into high frequency image data and low frequency image data. The high frequency image data is separated into windows, with each window containing pixels. The low frequency image data is separated into windows corresponding respectively to the plurality of windows of the high frequency image data, with each window containing pixels. For each window in the high frequency image data, a number of textured pixels in the window is determined, the textured pixels being pixels that vary greatly in luminance value with respect to other pixels in the window, and a window modification is determined such that the number of textured pixels in the window is reduced. For each corresponding window in the low frequency image data, the window modification is applied, and each textured pixel in the window is corrected based upon the other pixels in the window.