K.S. Sim, W.Z. Wan Ismail, L.W. Thong, W.K.Lim, and H.Y. Ting (Malaysia)
Contrast Enhancement, Histogram Equalization, Histogram Partition and Contrast Stretching
Histogram equalization (HE) is a conventional approach for image contrast enhancement. It gains popularity in parallel with other techniques such as unsharped masking and wavelet transform because of its simplicity and effectiveness in creating a good image. Nevertheless, HE technique is not desirable to be used in consumer electronics applications directly as it may cause effects, such as introducing undesirable artifacts, washed out appearance, and checkerboard effect due to the significant changes in image brightness. To surmount this drawback, we propose a new and robust technique, referred as Contrast-Stretching Dynamic Histogram-Equalization (CSDHE), which can simultaneously enhance visualization of the original image and preserve the brightness of the original image. It builds on the variational formulation of histogram equalization to define an optimal transformation function that is consisted of five steps. First, the input histogram is normalized, and then is smoothened by Gaussian filtering. The third step is break point detection process, followed by remapping of gray level allocation. In the final step, independent transformation function of each sub-histogram is constructed. The many experimental evaluations done on many images prove that this new technique can outperform the classical HE method and overcome the brightness-preserving problem. Consequently, very low contrast images can utilize this technique for enhancement.
Important Links:
Go Back
