| Paper Details: | Downloads: 1196 |
| Serial Number: | P1151442348
|
| Title: | Visual Perception Oriented CBIR envisaged through Fractals and Presence Score
|
| Authors: | Suhas S Rautmare and Anjali S Bhalchandra
|
| Abstract: | The objective of this research is to validate the feasibility of successful content based image search using fractals. Fractal based approach has been embedded into one of the theories of visual perception of an image. An important function of selective visual attention is to direct our gaze rapidly towards objects of interest in our visual environment. Two distinct pathways in human visual perception for scene recognition are what pathway and where pathway. They work simultaneously to perceive scene as a whole. The proposed approach simulates human visual perception mechanism similar to that of the brain. The Hausdorff Fractals have been used in what pathway and Presence Score has been used in where pathway to generate a composite feature vector. The feature vector captures color, shape, texture information and significant region of interest of the image through fractals and presence score. In order to improve retrieval performance in the sense of accuracy and time, similar images are clustered together through maxdistnace and maxclustersize based clustering approach. The results obtained with an optimum sized feature vector are good enough to validate the approach and the conceptualized analogy between human and computer based image processing. The search has also been successfully extended using text annotation to web search through Google to provide wider coverage of CBIR.
|
| Keywords: | What/Where Pathway, Hausdorff Fractals, Presence Score (PS), Segmentation, maxdistnace, maxclustersize, Visual Perception Oriented CBIR (VPO CBIR)
|
| Journal/Conference: | International Journal of Graphics, Vision and Image Processing
|
| Volume: | 14
|
| Issue: | 2
|
| Submission Date: | 10/16/2014 12:00:00 AM
|
| Review Date: | 11/9/2014 12:00:00 AM
|
| Publishing Date: | 12/9/2014 12:00:00 AM
|
| Article Downloads: | 1196
|
| Download: |
|