NEURAL NETWORK BASED IMAGE SEGMENTATION FORNATURAL IMAGES

Abstract

This paper presents a new Neural Network based region image segmentation using grow cut. This method is developed to obtain improved performance in seed based region growing method. The results shown in normal seeded region growing method are unsatisfactory so we propose a new method to increase the performance of the result by adding a neural network to the system. By training the neural network it will easily put the seeds for growing and merging. The performance of the proposed system is better compared with the seed region growing and the values inferred from result evaluation gives good result.