Abstract

The ability of the Human Visual System (HVS) to detect an object in an image is extremely fast and reliable but how can a machine vision system detects the salient regions? many algorithms have been proposed to solve this problem by extracting features in either spatial or spectral domain, in this paper, A novel saliency detection model is introduced by utilizing low level features obtained from Stationary Wavelet Transform domain. Here Stationary Wavelet Transform (SWT) is preferred as the wavelet transform than Discrete Wavelet Transform (DWT), Since DWT is not a time-invariant transform. So to make it translation invariant SWT is introduced. And also unlike the other wavelet transforms SWT does not require down sampling, So image size is same as original even after decomposition, thus there is no information loss in respective sub bands. Experimental results demonstrate that proposed model produces better performance by using SWT than by using DWT with the overall F-Measure value being high.