Abstract
Local Histogram Equalization (LHE), which is efficient for High Dynamic Range (HDR) compression, requires a large amount of calculation and induces unnatural image quality. Using checker-wise histogram acquisition and line buffer, memory accessing for local histograms is processed in O(1) time. For improving image quality, we proposed approximation of LHE by compositing gamma functions. Moreover the control of local histogram is developed by introducing mean values of bins for suppression of stripe artifacts arising from small number of bins and halo effect. Using C++, we achieved two million pixels per 0.45sec operation.
