Abstract: We have proposed image enhancement and segmentation based on fuzzy rule and graph cut method. A graph is constructed from the image using its intensity, texture and color profiles concurrently. A fuzzy rule based system is developed depending on the nature of image in order to find the weight which is required to develop feature for the specific image. The resulted weighted average of various image features is further used to frame normalized graph cuts. Further, the graph is bi-partitioned iteratively through the normalized graph cuts algorithm. As a result, we get optimum partitions which are the required segments of an image. We used the standard Berkeley segmentation database to test our algorithm and segmentation results are evaluated through index rand probabilistic, global consistency error, sensitivity, Dice similarity coefficient and positive predictive value. It is revealed that the proposed segmentation technique provides efficient results for different types of general images.
B. Basavaprasad and Ravindra S. Hegadi, 2017. A Fusion Technique of Image Enhancement and Segmentation using Fuzzy Rule and Graph Cut Method. Asian Journal of Information Technology, 16: 323-332.