적응적 함수를 이용한 히스토그램 기반의 Image Contrast 개선 방법

Abstract

In this paper, we propose an algorithm for contrast enhancement based on adaptive histogram equalization to improve image quality. Excessive contrast enhancement is the main problem on histogram based enhancement techniques because it turns out to be undesired resultant images ranging from noise amplification to loss of visual information. In order to overcome this, we propose two adaptive functions and adaptive clipping-limit. The clipping-limit is usually set to fixed-value to limit the maximum enhancement. However, since images have the varying characteristics, our algorithm creates adaptive clipping-limit by based on image Entropy. Also, to modify input histogram under the maximum pixel number limitation, mathematically driven two adaptive functions are proposed. Both functions decrease input histogram under the adaptive clipping-limit and compensate clipped pixels to the modified histogram. While computing the adaptive functions, the shapes of input histogram are completely preserved. Thus, the contrast of input images can be adaptively improved in stochastically. Our proposed method displays improved PSNR average values up to 55% than the conventional methods and illustrations more enhanced visual image contrast in the simulation results.