Local figure–ground cues are valid for natural images

Abstract

Figure-ground organization refers to the visual perception that a contour separating two regions belongs to one of the regions. Recent studies have found neural correlates of figure-ground assignment in V2 as early as 10-25 ms after response onset, providing strong support for the role of local bottom-up processing. How much information about figure-ground assignment is available from locally computed cues? Using a large collection of natural images, in which neighboring regions were assigned a figure-ground relation by human observers, we quantified the extent to which figural regions locally tend to be smaller, more convex, and lie below ground regions. Our results suggest that these Gestalt cues are ecologically valid, and we quantify their relative power. We have also developed a simple bottom-up computational model of figure-ground assignment that takes image contours as input. Using parameters fit to natural image statistics, the model is capable of matching human-level performance when scene context limited.

Simultaneous Segmentation and Figure/Ground Organization Using Angular Embedding

“…Historically, more attention has been paid to segmentation, though some important studies of figure/ground exist, focusing on contour and junction structure [13,11,25,32] or specific cues [10] such as convexity [21] or lower-region [29]. Recent work has revived interest on figure/ground discrimination [24,16] and the related problem of depth ordering [15,26].…”

“…Leichter and Lindenbaum take the human-drawn ground-truth segmentations [19] and human figure/ground annotations [10] of the BSDS images and learn a conditional random field (CRF) for assigning boundary ownership. They use curve and junction potentials, exploiting convexity, lower-region, fold/cut, and parallelism cues.…”

“…It is also partially motivated by convenience, as it allows us to train our figure/ground classifier on the same dataset, the BSDS, as our segmentation algorithm, due to the availability of pre-existing annotations [19,10]. Consequently, our work is not directly comparable to that for which depth ordering or three-dimensional reconstruction is the ultimate goal [15,26].…”

Dimensionality reduction by minimizing nearest-neighbor classification error

“…linear and supervised, namely LDA, MFA, LSDA, SLPP, NDA, NCA and LMNN all of them with and without a PCA preprocessing. For LDA our own implementation was used, however for the rest, we used freely available implementations from the authors of Cai et al (2007); Bressan and Vitrià (2003); Weinberger et al (2006);Fowlkes et al (2007). For each of the baseline methods, the corresponding algorithm parameters were properly adjusted, and only the best result obtained in each case is shown.…”

Automatic Highly Accurate Estimation of Gaussian Noise Level in Digital Images Using Filtration and Edges Detection Methods

“…Then there is a process of specifying the ROI and its corresponding standard deviation σ h (for image f h ) in the cycle with the counter t. For each iteration t the threshold T h (17) is calculated. Based on the threshold T h and the averaged image of the noise level f dc , the ROI is calculated where the function ψ 1 is described by the formula (13). If the σ h change for the iteration t in reference to the previous value…”