Saliency Detection Based on Multiscale Extrema of Local Perceptual Color Differences

Abstract

Visual saliency detection is a useful technique for predicting, which regions humans will tend to gaze upon in any given image. Over the last several decades, numerous algorithms for automatic saliency detection have been proposed and shown to work well on both synthetic and natural images. However, two key challenges remain largely unaddressed: 1) How to improve the relatively low predictive performance for images that contain large objects and 2) how to perform saliency detection on a wider variety of images from various categories without training. In this paper, we propose a new saliency detection algorithm that addresses these challenges. Our model first detects potentially salient regions based on multiscale extrema of local perceived color differences measured in the CIELAB color space. These extrema are highly effective for estimating the locations, sizes, and saliency levels of candidate regions. The local saliency candidates are further refined via two global extrema-based features, and then a Gaussian mixture is used to generate the final saliency map. Experimental validation on the extensive CAT2000 data set demonstrates that our proposed method either outperforms or is highly competitive with prior approaches, and can perform well across different categories and object sizes, while remaining training-free.

Colour key-point detection

“…A colour adaptation from Ishikura et al [9] extracts keypoints marginally in CIELAB colour space. Rassem and Khoo [10] measure quantified colour histogram to extract key-point.…”

Saliency detection based on structural dissimilarity induced by image quality assessment model

“…The center-surround absolute difference on the biologically motivated fea-tures was first proposed by Itti et al 6 and is widely used to detect the local distinctive regions. 7,8 Later, by proposing a graph model to globally integrate the image regions, Harel et al showed that the global distinctiveness could better predict saliency. Further, the spatially weighted global absolute difference of image patches in the principal components space 9 or CIELab color space 3,10 also presents high accuracy in saliency detection.…”

“…This paper contributes toward saliency detection by exploiting IQA models. Many previous saliency models computed the absolute difference of features to measure the distinctiveness of image regions in the local 6,8 or global extent. 3,9,30 However, according to the studies on IQA, the structural similarity is validated as a better similarity measure than the absolute difference in the context of HVS perception.…”

“…To compute the structural dissimilarity between image patches, I is first decomposed into a set of structural features. According to studies, such asRefs 6,8,58, both the luminance and chrominance information are crucial for saliency detection. Therefore, structural…”

Tensor Decomposition for Salient Object Detection in Images

“…Then, a region-based contrast was proposed to enhance the spatial information on the final saliency map. Ishikura et al [14] proposed an unsupervised saliency detection method based on multi-scale extrema of local perceptual color difference. However, their method was only used to estimate the locations, sizes, and saliency levels of candidate regions instead of the pixel-level saliency map.…”