液晶与显示, 2016, 31 (1): 117, 网络出版: 2016-03-22
基于区域特征融合的RGBD显著目标检测
RGBD salient object detection based on regional feature integration
目标检测 深度信息 区域特征 随机森林 监督学习 object detection depth information regional feature random forest supervised learning
摘要
为了对各类自然场景中的显著目标进行检测, 本文提出了一种将图像的深度信息引入区域显著性计算的方法,用于目标检测。首先对图像进行多尺度分割得到若干区域, 然后对区域多类特征学习构建回归随机森林, 采用监督学习的方法赋予每个区域特征显著值, 最后采用最小二乘法对多尺度的显著值融合, 得到最终的显著图。实验结果表明, 本文算法能较准确地定位RGBD图像库中每幅图的显著目标。
Abstract
In order to detect all kinds of salient object in nature scene, we present a detection method which adds regional depth information of image to saliency computation, to apply to object detection. Firstly, we get several region through multi-level image segmentation. Then, we build the regression random forest by learning varieties of regional features, and use the supervised learning approach to map the regional feature vector to a saliency score. Finally, we fuse the saliency scores across multiple levels by least square method, yielding the saliency map. Experiments show our method can accurately locate the salient objects from RGBD images.
杜杰, 吴谨, 朱磊. 基于区域特征融合的RGBD显著目标检测[J]. 液晶与显示, 2016, 31(1): 117. DU Jie, WU Jin, ZHU Lei. RGBD salient object detection based on regional feature integration[J]. Chinese Journal of Liquid Crystals and Displays, 2016, 31(1): 117.