Abstract:
Content-based image retrieval (CBIR) is widely used in various todays application; including web-based image retrieval, medical, security, etc. However, the most research topic is hot to reduce semantic gap which is the different between user specify input and the queried output images.
To solve the problem, ones can develop a method to extract meaningful features from images or give more details to the particular pictures for later classification process. In this research, we investigate the use of fractal geometry analysis to compute the complexity of images as a feature for CBIR system. Different fractal estimators; Box Counting, Higuchi and Scaling Properties of Variance (SPV), have been adopted to compare the computation efficiency and the effectiveness amount the other.
In conclusion, this research aims to develop a method to directly extract relevant information from an input image for CBIR system. Moreover, the proposed method intent to be used for a huge image database in which high retrieval rate is required.