A Novel Feature Extraction Method Based on Segmentation over Edge Field for Multimedia Indexing and Retrieval

In this work we focused on visual content extraction. Digital images have some basic features, such as texture, colour and shape. Through low-level bit processing, and using colour intensity changes across pixels, we try to achieve some form of mid-level meaningful description of the image. Starting from the well-known Canny Edge algorithm we convert the image map to an edge map. From the edges, and performing analysis based on perceptual rules, we achieve a segmentation of the image, which is proved sound for content-based indexing and retrieval purposes.