Saliency estimation using a non-parametric low-level vision model

Abstract

Many successful models for predicting attention in a scene involve three main steps: convolution with a set of filters, a center-surround mechanism and spatial pooling to construct a saliency map. However, integrating spatial information and justifying the choice of various parameter values remain open problems. In this paper we show that an efficient model of color appearance in human vision, which contains a principled selection of parameters as well as an innate spatial pooling mechanism, can be generalized to obtain a saliency model that outperforms state-of-the-art models.Scale integration is achieved by an inverse wavelet transform over the set of scale-weighted center-surround responses. The scale-weighting function (termed EC SF ) has been optimized to better replicate psychophysical data on color appearance, and the appropriate sizes of the centersurround inhibition windows have been adjusted by training a Gaussian Mixture Model on eye-fixation data, thus avoiding ad-hoc parameter selection. Additionally, we conclude that the extension of a color appearance model to saliency estimation adds to the evidence for a common low-level visual front-end for different visual tasks.

Joint Multi-Image Saliency Analysis for Region of Interest Detection in Optical Multispectral Remote Sensing Images

“…We compare our model with eight competing models through qualitative and quantitative experiments. The eight saliency detectors are those from Itti et al [11], Achanta et al [12], Achanta et al [13], Jiang et al [24], Murray et al [22], Goferman et al [23],İmamoglu et al [9], and Zhang et al [18], herein referred to as ITTI, FT, MSSS, CB, SIM, Context-Aware model (CA), TMM, and FDA-SRD, respectively. These eight models are selected for the following reasons: high citation rate (the classical ITTI models), recency (CA, TMM and FDA-SRD), variety (ITTI is biologically motivated, the FT and MSSS models estimate saliency in frequency domain, and SIM is a top-down method), and affinity (CB operates in superpixels and FDA-SRD is related to remote sensing image processing).…”

“…Many recent studies are inspired by Itti's biologically based idea. Murray et al [22] introduced saliency by induction mechanisms (SIM) based on a low-level vision system. A reduction in ad hoc parameters was achieved by establishing training steps for both color appearance and eye-fixation psychophysical data.…”

Fast Binary Coding for the Scene Classification of High-Resolution Remote Sensing Imagery

“…However, it remains unclear how this strategy can act on the performance in other classification framework. In contrast with [39], where only one saliency detection method (context-aware) is evaluated, we investigate fifteen kinds of typical methods, which are AC [53], attention by information maximization (AIM) [54], context-aware (CA) [52], context-based (CB) [55], discriminative regional feature integration (DRFI) [56], frequency-tuned (FT) [57], graph-based visual saliency (GBVS) [58], IM [59], low rank matrix recovery (LRR) [60], maximum symmetric surround (MSS) [61], rarity-based saliency (RARE) [62], salient segmentation (SEG) [63], self-resemblance (SeR) [64], spectral residual (SR) [65] and saliency using natural statistics (SUN) [66], and present detailed comparative experiments for them. We do not review these methods due to the limited space, and details of each method can be found in its corresponding paper.…”

Saliency Detection Using Entropy Weight and Weber's Law

“…이 방법을 토대로, 위상 스펙트럼이 세일리 언시를 검출하기 위한 주된 요소라 판단하고 푸리에 변 환의 위상에 기반 한 세일리언시 검출 방법이 발표되었 다 [11] . 웨이블릿 계수로부터 국부 및 전역 대비 도를 계 산하고, 이를 이용한 세일리언시 검출 방법 [12] 이 제안되 었으며, 다중 스케일 특성에 대한 역 웨이블릿 변환을 이용한 방법 [13] [14] .…”