Learn More
In this paper, we propose a geometric framework for stop sign detection based on polylines. We first propose a scheme for the extraction of 1-piece and 2-piece polylines from connected components of edge pixels. This scheme contains three basic steps: i) dominant point extraction, ii) linearity verification, and iii) partitioning. We then propose a(More)
Rectangular shape detection has a wide range of applications, such as license plate detection, vehicle detection and building detection. In this paper, we propose a robust framework for rectangular shape detection based on the channel-scale space of RGB images. The framework consists of algorithms developed to address two issues of a shape attention (i.e.,(More)
Background subtraction is a fundamental and critical component in many high-level video processing applications. It remains a challenging task in complex scenes due to difficult factors such as illumination variations, cast shadows and dynamic backgrounds. In this work, we propose a novel background subtraction method for coping with RGBD videos. Our method(More)
Concave region partitioning is valuable for object modeling and recognition. A key issue to the design of an efficient partitioning method is on the selection of cut points. In this paper, we propose to use imbalanced points in a region to characterize cut points in the contour of the region, motivated by the good corner nature of imbalanced points.(More)
Content-aware image scaling methods have received great attentions in the past 10 years. Most of them are proposed only for monocular 2D images. With the development of 3D display techniques, it is very desirable to extend these methods on stereoscopic images. The existing stereo methods fail in preserving objects with straight lines. In this paper, we(More)
Connected components of edge pixels are valuable for many applications, e.g., Image segmentation, contour extraction, and text detection. However, a connected component of edge pixels may not form a simple curve due to the existence of bifurcation pixels, and thus limits its application in shape modeling. In this paper, we propose an approach to remove(More)
Images consist of different low-level features, such as Points of Interest (POIs) and Regions of Interest (ROIs). A distinction between POIs and ROIs is that the latter ones have intrinsic scale information while the former ones may not have. In this paper, we propose a scheme to integrate these two kinds of image features. The proposed scheme optimizes(More)
Automatic localization of gene expression regions in Drosophila embryonic images is an important step in a computational system for the discovery of patterns of gene-gene interaction. In this paper, we propose a thresholding framework that selects an optimal threshold via a consistency criterion on good candidate thresholds provided by multiple response(More)