摘要
针对俯视图像中往往存在反光区域使得导航算法性能下降这一问题,提出一种反光区域检测与分割算法.在HSV颜色空间中对图像反光区域的颜色特征进行分析,发现反光区域具有图像饱和度分量较小、亮度分量较大、从反光区域的中心位置到边缘区域亮度逐渐变低的特点,利用该特征实现了对图像中反光区域的检测和分割,并通过实验对算法性能进行测试.结果表明,该算法可以准确地检测出图像中的反光区域.
In lighting of the problem about illuminant reflection regions being found in a birdview image and it leading to performance degradation of the navigation system, an algorithm of illuminant reflection detection and segmentation is proposed. The illuminant reflection regions in a birdview image are analyzed in HSV color space. It is found that the saturation components of the image regions are small while brightness components are large and brightness values decrease from central positions to marginal positions. And then illuminant reflection is detected and segmented based on these image features. Finally, experiments are run to testify the algorithm, and the experimental results show that the performance of algorithm is reasonable.
出处
《天津工业大学学报》
CAS
北大核心
2013年第6期59-62,共4页
Journal of Tiangong University
基金
天津市科技支撑计划重点资助项目(12ZCZDGX04200)
天津市高等学校科技发展基金计划项目(20110707)
关键词
视觉导航
地图构建
反光检测
HSV颜色空间
vision-based navigation
map building
illuminant reflections detection
HSV color space
