Improving visual presentation of bas relief image from historical temple

Abstract

Two stages are developed in this research namely optimal pre-processing and main-processing. The pre-processing is to enhance the presentation of image relief by using contrast stretching, histogram equalization and adaptive histogram equalization. Here the presentation of image is vital for the input of main process. Three operations of image enhancement are employed to improve image contrast, including contrast stretching, histogram equalization and adaptive histogram equalization. Comparison between contrast stretching with a value 1.1 and histogram equalization show that the gray-level of histogram equalization result are distributed. Experiment of show that the adaptive histogram equalization has produced the best image enhancement. After doing image enhancement the next step is to do threshold and edge detection operations. This procedure enhances horizontal or vertical edge of the figure. The image enhancement operation is very useful to remove the illumination no uniformities. Threshold process using Otsu algorithm shows that both of LPF kernel 1 and 2 output image have a similar result rather than LPF kernel 3. Three of LPF kernels has produced two different threshold values. Kapur algorithm show that both of LPF kernel 2 dan 3 output image have a similar result of rather than LPF kernel 1 output image. The experiment also show that Sobel and Canny operators has enhanced edges and weaken weak textures.