Automatic Contrast Enhancement Using Pixel-Based Calibrating and Mean Shift Clustering
In this paper, we present the method for automatic contrast enhancement of color image. The base concept of method is that an image has its own reference luminance level and each pixel has its own characteristic luminance that is brighter or darker than reference luminance level. In the proposed method, a given color image is converted to HSV color space from RGB color space firstly. Next, each pixel in the image find out the own characteristic luminance based on the reference luminance level. The characteristic luminance is calibrated to the target luminance that will get the acceptable luminance. We apply alpha blending the original luminance and characteristic luminance to reduce the HALO artifact and preserve details of darker area by mean shift clustering.
KeywordsColor Image Dark Area Luminance Component Target Luminance Adaptive Histogram Equalization
Unable to display preview. Download preview PDF.
- 8.Jin, Y., Fayad, L., Laine, A.: Contrast enhancement by multi-scale adaptive histogram equalization. In: Proc. SPIE, vol. 4478, pp. 206–213 (2001)Google Scholar