Boundary Detection Using Double-Opponency and Spatial Sparseness Constraint

Abstract

Brightness and color are two basic visual features integrated by the human visual system (HVS) to gain a better understanding of color natural scenes. Aiming to combine these two cues to maximize the reliability of boundary detection in natural scenes, we propose a new framework based on the color-opponent mechanisms of a certain type of color-sensitive double-opponent (DO) cells in the primary visual cortex (V1) of HVS. This type of DO cells has oriented receptive field with both chromatically and spatially opponent structure. The proposed framework is a feedforward hierarchical model, which has direct counterpart to the color-opponent mechanisms involved in from the retina to V1. In addition, we employ the spatial sparseness constraint (SSC) of neural responses to further suppress the unwanted edges of texture elements. Experimental results show that the DO cells we modeled can flexibly capture both the structured chromatic and achromatic boundaries of salient objects in complex scenes when the cone inputs to DO cells are unbalanced. Meanwhile, the SSC operator further improves the performance by suppressing redundant texture edges. With competitive contour detection accuracy, the proposed model has the additional advantage of quite simple implementation with low computational cost.

A Biologically-Inspired Framework for Contour Detection Using Superpixel-Based Candidates and Hierarchical Visual Cues

“…The mean computation time to compute one contour map with gPb [ 1 ] is 191 s, while our algorithms only takes 13 s (the computer used here is Intel Core 2, 2.5 GHZ with 8.0 G RAM). Our method is a littler slower than CO [ 48 ] and SCO [ 75 ]. Note that, CO [ 48 ] and SCO [ 75 ] only perform at a single scale, because their extension into multi-scale space did not yet produce clear performance improvements [ 48 , 75 ].…”

“…Our method is a littler slower than CO [ 48 ] and SCO [ 75 ]. Note that, CO [ 48 ] and SCO [ 75 ] only perform at a single scale, because their extension into multi-scale space did not yet produce clear performance improvements [ 48 , 75 ]. The computation time of our method is promising as a multi-scale method.…”

Superpixel Segmentation Based on Anisotropic Edge Strength

“…com/ASD) [44], and the neuronal structures in electron microscopy stacks dataset (NSEMS; http://brainiac2.mit.edu/isbi_challenge/home) [45]. In addition, in order to explore the impact of different kinds of edge strength on the segmentation accuracy, we also tested the SH methods that incorporate edge strength obtained by the structured forest edge (SH+SFE) method [28], the sparseness-constrained color-opponency (SH+SCO) method [26], the automated anisotropic Gaussian kernel (SH+AAGK) method [19], and the surrounded-modulation edge detection (SH+SED) method [27]. Furthermore, we also selected the widely used SLIC method [8] for comparison.…”

Image processing of concentrated and scattered objects

“…The angle between the rows of zeros in the mask turns out to be the angle of the gradient vector. The mask allows us to obtain the maximum gradient value [19].…”