Image segmentation is one of the important tasks in modern era. Traditional image segmentation techniques make use of conventional methods of clustering which suffers from problems. First problem is over segmentation which degrades the quality of the image. Moreover, parameter estimation is a big problem in traditional segmentation techniques. A new technique called JSEG is proposed towards this goal. This approach does not tend to apply to a specific model. Further some; it is used to test the homogeneity of the color pattern in regions in images. In this paper, JSEG along with enhanced clustering method called Modified K means clustering is presented. It has been investigated that the method is quite computationally feasible. Matlab software is used for simulation along with image processing toolbox