光学学报, 2018, 38 (3): 0315002, 网络出版: 2018-03-20
基于双目视觉的显著性目标检测方法 下载: 1071次
Salient Object Detection Method Based on Binocular Vision
图像处理 目标检测 双目视觉 多特征融合 区域分割 视觉显著性 image processing object detection binocular vision multi-feature fusion region segmentation visual saliency
摘要
针对现有的显著性目标检测算法在受到相似背景干扰时, 易出现目标检测准确度低、稳定性差的问题, 提出一种基于双目视觉的显著性目标检测方法。受人眼视觉特性启发, 将双目视觉模型感知的深度信息作为显著性特征与多特征聚类分割结果进行协同处理, 定量分析图像区域级的深度显著性, 再将全局显著性与区域深度显著性进行加权融合, 突出目标区域, 根据融合结果的区域分布进行背景抑制, 完成显著性目标的检测。实验结果表明, 与现有的显著性目标检测算法相比, 该算法有效地抑制了相似背景的干扰, 并且准确度高、稳定性好。
Abstract
Aiming at the problem that the existing salient object detection algorithms suffers from the similar background interference, the detection accuracy of the target is low and the stability is poor. We propose a salient object detection method based on binocular vision. Firstly, inspired by the visual characteristics of the human eye, we consider the depth information acquired by binocular vision model as the salient features based on human visual characteristics. Secondly, we use the depth information and the result of region segmentation based on multi-feature fusion clustering to analyze the regional level depth saliency of image quantitatively. Thirdly, we make the weighted fusion of the global saliency map and regional level depth saliency map to highlight the objection area. Finally, we suppress the background to complete salient object detection based on the regional distribution of fusion results. The results show that compared with the existing methods, the proposed method can effectively suppress the interference of similar background with high accuracy and stability simultaneously.
李庆武, 周亚琴, 马云鹏, 邢俊, 许金鑫. 基于双目视觉的显著性目标检测方法[J]. 光学学报, 2018, 38(3): 0315002. Li Qingwu, Zhou Yaqin, Ma Yunpeng, Xing Jun, Xu Jinxin. Salient Object Detection Method Based on Binocular Vision[J]. Acta Optica Sinica, 2018, 38(3): 0315002.