Abstract
Salient object detection (SOD) has been attracting a lot of interest, and recently many computational models have been developed. In this paper, we formulate a SOD model, in which saliency map is computed as a combination of the colour, its distribution-based saliency and orientation saliency. Similar to traditional SODs, the proposed method is based on super-pixel segmentation and super-pixelutilizes both colour and its distribution-based saliency to generate a coarse saliency map. However, distinct from traditional SODs, we further use orientation contrast to optimize the coarse saliency map to obtain an improved saliency map. Our contributions are twofold. First, we combine colour uniqueness and its distribution with local orientation information (LOI) used in Itti’s model to effectively improve profiles of salient regions. Second, a reciprocal function is defined to substitute the Gabor function used in LOI and we have proved that the substitution could detect relatively homogeneous and uniform regions at the boundary of salient object, whereas it is what the traditional models lack. Our approach significantly outperforms state-of-the-art methods on four benchmark datasets while, we demonstrate that the proposed method runs as fast as most existing algorithms.
Original language | English |
---|---|
Pages (from-to) | 1616-1624 |
Number of pages | 9 |
Journal | IET Image Processing |
Volume | 13 |
Issue number | 10 |
Early online date | 18 Apr 2019 |
DOIs | |
Publication status | E-pub ahead of print - 18 Apr 2019 |
Keywords
- feature extraction
- image segmentation
- object detection
- image resolution
- image colour analysis
- filtering theory