摘要
结合RGB图像和深度图像,提出了一种新的基于深度学习的无监督物体识别算法KSAE-SPMP。采用标准的RGB-D数据库2D3D来验证新提出的算法。实验结果表明,与之前提出的基于RGB-D的物体识别算法相比,KSAE-SPMP算法取得了最高的识别准确率,此算法能够很好地完成RGB-D物体的识别。
Combined with RGB and depth images, a novel unsupervised object recognition algorithm KSAE-SPMP based on deep learning was put forward. A standard RGB-D database 2D3D was adopted to verify the proposed algorithm. Experimental results demonstrated that compared with RGB-D algorithm based on object recognition proposed previously, KSAE-SPMP algorithm has the highest accurate identification rate, which is able to complete the RGB-D object recognition commendably.
出处
《移动通信》
2015年第10期52-56,共5页
Mobile Communications
基金
浙江省移动网络应用技术重点实验室(2010E10005)
浙江省新一代移动互联网用户端软件科技创新团队(2010R50009)
新型输入引擎及搜索与识别算法研究(2012R10009-19)
浙江省重中之重学科开放基金项目(xkxl1305)