Optimized Entropy Based Image Segmentation
Subhaluxmi Sahoo, Sony Snigdha Sahoo, Subham Kumar, Tashmin Mishra, Kishan Kumar Singh · 2020 International Conference on Computer Science, Engineering and Applications (ICCSEA) · 2020
Segmentation performs a very important role in image processing and computer vision. The automatic thresholding methods are very popular because of their ease of implementation. Depending upon the nature of histogram, thresholding can be used to divide an image into many classes. The decision of threshold value is slightly difficult and has to be done keeping in view the maximization of certain metric. Here in this paper, we have used two entropies, namely Tsallis entropy and Renyi’s entropy and we have combined them linearly and maximized this entropic function using genetic algorithm. We have applied this computation over certain number of sub windows in the image. We have seen that this method is useful for segmenting images taken in low light conditions. We have compared our proposal against some standard methods and found out that it performs better both qualitatively and quantitatively.