Image segmentation by graph partitioning

In this paper we propose an hybrid method for the image segmentation which combines the edge-based, region-based and the morphological techniques in conjunction through the spectral based clustering approach. An initial partitioning of the image into atomic regions is set by applying a watershed method to the image gradient magnitude. This initial partition is the input to a computationally efficient region segmentation process which produces the final segmentation. We have applied our approach on several images of the Berkeley Segmentation Dataset. The results reveal the accuracy of the propose method.