| Paper Details: | Downloads: 993 |
| Serial Number: | P1151404314
|
| Title: | Bench Marking Higuchi Fractal for CBIR
|
| Authors: | Suhas Rautmare and Anjali Bhalchandra
|
| Abstract: | Content Based Image Retrieval (CBIR) has been approached through color, shape, texture and many other approaches. CBIR can also be approached through fractals. Fractal dimension based approach and fractal compression based approach are two variants in this domain. The proposed work implements a fractal dimension based approach supported by morphological operations to enhance the results. The images under consideration are resized to 128x128. Fractals are patterns that exhibit self similarity. The paper discusses Higuchi Fractal dimension (HFD) based approach for CBIR. HFD provides direction dependent and direction independent analysis of the image. To enhance the results, morphological operations have been performed. The main purpose of adopting morphological operations is to extract image components that are useful in the representation and description of the content of image. Morphological operations like fill Holes, clear border object and dilation of the image have been used to enhance the results. A feature vector of the size 256, 128 for columns and 128 for rows, has been used to identify the relevant match of the images through Euclidean distance measure. The paper establishes a bench mark for Higuchi fractal dimension to be used for CBIR. The performance of the proposed system is good enough in comparison with other fractal based systems to be considered as a bench mark for using Higuchi fractal in CBIR process. The results and analysis has been presented for the proposed approach.
|
| Keywords: | Content Based Image Retrieval (CBIR), Fill Holes, Higuchi Fractal Dimension (HFD), Morphological Operation, Precision, Recall
|
| Journal/Conference: | International Journal of Graphics, Vision and Image Processing
|
| Volume: | 14
|
| Issue: | 1
|
| Submission Date: | 1/21/2014 12:00:00 AM
|
| Review Date: | 1/31/2014 12:00:00 AM
|
| Publishing Date: | 2/17/2014 12:00:00 AM
|
| Article Downloads: | 993
|
| Download: |
|