An abundance of various automated and semi-automated segmentation techniques can be found in literature that caters to wide range of image analysis and understanding applications. To cater such a wide coverage of algorithms, we have proposed Multi- Faceted Hierarchical Image Segmentation Taxonomy (MFHIST). We also proposed a computationally inexpensive generic hybrid color-texture feature integration methodology applied to natural color image segmentation. Automatic and proper segmentation along with texture characterization generates informative knowledge representations, which inspired the present study, and hence have come up with a framework for providing meaningful representation of the given image objectively. The proposed framework is expected to be useful in the field of computer vision for cases, where prior knowledge about the scene is unknown.
Research interests:
Image Processing, Computer Vision, Machine Learning, Remote Sensing