摘要
天际线检测在机器视觉导航、地理位置定位、船只检测和港口安全、森林防护中有重要的作用。首先采用区域协方差算法RCA(Regional Covariance Algorithm)将图像分割成天空与非天空区域,实现天际线检测区域的粗提取;接着提出一种基于分割粗区域全局梯度均值与局部梯度均值融合算法在训练样本上确定梯度阈值,并以此算法检测出天空与非天空区域的起始点;然后以起始点开始在分割的非天空区域采用相邻梯度极大值AGM(Adjacent Gradient Maxima)算法搜索最大可能天际线存在的像素点位置坐标,最终达到检测出输入图像中的天际线。该算法在内华达大学机器视觉实验室Web Set数据集进行了测试,实验结果表明:该算法能有效地检测出天空与非天空区域的分界线,具有良好的有效性和时效性。
Skyline detection plays an important role in machine visual navigation, geographical position location, ship detection and port security and forest protection. Firstly, the regional covariance algorithm （RCA） is used to divide the image into the sky and the non-sky regions. Then, a new method based on the combination of the global gradient mean and the local gradient mean value is proposed to determine the starting point of the sky and the non-sky regions. And then, from the starting point, the maximum possible skyline is detected by using the adjacent gradient maxima（AGM） algorithm in the segmentation of the non-sky regions. The proposed algorithm is tested on the Web Set data set in University of Nevada＇ s Machine Vision Laboratory. The experimental results show that the algorithm can effectively detect the boundary between the sky and the non-sky regions, and has good validity and timeliness.
出处
《计算机应用与软件》
CSCD
2016年第9期176-179,220,共5页
Computer Applications and Software
基金
国家自然科学基金项目(61201435)
湖南省高校科技创新团队支持项目(湘教通[2012]318号)
关键词
天际线
视觉导航
区域协方差
相邻梯度
Skyline Visual navigation Regional covariance Adjacent gradient
