Discriminative Generative Contour Detection

• DOI: 10.5244/c.27.74
• Publication Date: 2014-01-10T09:21:58Z
• Authors: Chao Zhang; Xiong Li; Xiang Ruan; Yuming Zhao; Ming-Hsuan Yang
• Keywords: Discriminative model; Generative model; Active contour model

Contour detection is an important and fundamental problem in computer vision which finds numerous applications.Despite significant progress has been made the past decades, contour from natural images remains a challenging task due to difficulty of clearly distinguishing between edges objects surrounding backgrounds.To address this problem, we first capture multi-scale features pixel-level segmentlevel using local global information.These are mapped space where discriminative information captured by computing posterior divergence Gaussian mixture models then used train random forest classifier for detection.We evaluate proposed algorithm against leading methods literature on Berkeley segmentation Weizmann horse data sets.Experimental results demonstrate that performs favorably state-of-the-art terms speed accuracy.