As a regional contrast enhancement method, adaptive histogram equalization (AHE) has been attracting more attentions in particular for medical imaging and photo editing since few decades ago, because it rates high importance on details and confidence of the enhanced contents for either observation or system reliability, and high dynamic range (HDR) purpose. Yet, currently simplified methods cannot fully address the high complexity O(M^2 ) issue of the AHE. In this work, a method termed integral image based AHE (iAHE) realizes the complexity with simply O(1) while maintaining an identical image quality as that of AHE, and it is also the simplest approach to achieve this comparison with all existing methods. The strategy behind this is to omit the “required” calculation of the pdf and exploit a shared cumulative information for the cdf. To prove the effectiveness of the proposed scheme, existing simplified techniques are involved in comparison for a comprehensive demonstration. Although the proposed iAHE is slower than some of the former schemes by around an order of runtime, they inevitably accompany either degraded regional contrast or serious artifact. All of the results show that the proposed iAHE can provide a great practical value to the existing commercial products which have applied generalized AHE such as digital cameras, and it is also eligible for medical imaging, ultrasound imaging, computer vision, and object detection applications.
|
技術摘要：
專利行銷平台
