Identification and Ranking of Relevant Image Content

The work in this thesis proposes an image understanding algorithm for automatically identifying and ranking different image regions into several levels of importance. Given a color image, specialized maps for classifying image content namely: weighted similarity, weighted homogeneity, image contrast and memory color maps are generated and combined to provide a perceptual importance map. Further analysis of this map yields a region ranking map which sorts the image content into different levels of significance. The algorithm was tested on a large database that contains a variety of color images. Those images were acquired from the Berkeley segmentation dataset as well as internal images. Experimental results show that our technique matches human manual ranking with 90% efficiency. Applications of the proposed algorithm include image rendering, classification, indexing and retrieval. Adaptive compression and camera auto-focus are other potential applications