A new approach of gradient and threshold based histogram equalization for contrast enhancement

This paper presents a new histogram equalization technique, based on gradient and threshold values. The background pixels of the image are associated with low gradient intensities and the detailed pixels are associated with high gradient intensities. The threshold value is usually used for segmentation. In the proposed method, at first the image has been divided into two regions according to the ascending order of gradient intensities. Then each region has also been divided into two subsections according to the corresponding threshold value. Therefore, four subsections have been obtained from the original image. After that, two sub-images have been formed from the subsections which are associated with the background of the image and the subsections which are associated with the detailed regions. Then classical HE technique has been applied in each sub-image and the enhanced image has been obtained by composing the processed sub-images. Experimental results of the proposed method have been compared with other two stat-of-arts methods. From the experimental results, it can be seen that the presented method provides better results compared to other methods. The proposed method preserves the brightness of the image with natural looks by giving low values of average mean brightness error (AMBE). The presented method also maintains the quality of the image by giving high values of peak signal to noise ratio (PSNR) and low values of mean square error (MSE).