Improved Techniques for Unsupervised Image Segmentation
Li Fu Gao, Jie Xia, Junli Liang, Shuyuan Yang · 2006
This paper proposed a novel automatic image-segmentation system which is an improvement to the marker-based watershed transform. A new marker-extracted approach was proposed to extract the regional minima from the low frequency components in the gradients map. The extracted minima constitute the binary marker image. Then the markers are imposed on the original gradients as its minima, and suppress its all intrinsic minima. Finally, the watershed algorithm is applied to the modified gradients to perform the segmentation. Across a variety of image types, it is proven that this new system can obtain meaningful and homogeneous regions with accurate, consecutive and one-pixel wide boundary. Additionally compared with other methods, this system requires fewer computations and simpler parameters and can more efficiently reduce the over-segmentation of the watershed algorithm