摘要
提出一种新的无人机监控图像实时目标识别算法。首先将获取的无人机监控图像应用自适应阈值分割将其转换为二值图像。对二值图像进行形态学处理,定位潜在目标出现的位置。最后对潜在目标区域再次应用局部自适应阈值分割获取目标,同时给出每个目标的图像坐标位置。飞行试验表明该算法保证实时性的情况下,有较高的识别正确率。
In this paper,we propose a new real-time object recognition algorithm for aerial surveillance video.The proposed algorithm first segments the captured video using the adaptive threshold segmentation method to get the binary image.Using morphological image pro- cessing and edge information we can position the potential object areas.Finally,we again use the adaptive threshold segmentation method for the potential object areas to get the final result,but also give the position of each object in the image.Experimental results using video captured from our mini-Unmanned Aerial Vehicle (UAV) show that our proposed real-time object recognition algorithm is both effective and efficient.
出处
《遥测遥控》
2007年第2期1-4,共4页
Journal of Telemetry,Tracking and Command
关键词
自适应阈值分割
形态学图像处理
无人机
Adaptive threshold segmentation
Morphological image processing
UAV
