现代电子技术2025，Vol.48Issue(13)：20-28,9.DOI:10.16652/j.issn.1004-373x.2025.13.003
面向复杂场景目标提取的颜色增强与连通区域构造方法
Color enhancement and connected region construction method for object extraction in complex scenes
李明禄 1赵春青 2侯茂新 2王肖霞 1杨风暴1
作者信息
- 1. 中北大学 信息与通信工程学院,山西 太原 030051
- 2. 中兵智能创新研究院有限公司 群体协同与自主实验室,北京 100072
- 折叠
摘要
Abstract
It is difficult to effectively extract the complete object region in complex scenes by the existing feature extraction methods,so a method of color enhancement and connected region construction based on HSV space nonlinear is proposed.On the basis of the object color features,the histogram distribution characteristics are analyzed to obtain the multilevel thresholds for object extraction.For the case of poor object extraction due to different lighting conditions,the enhanced extraction is performed by constructing the nonlinear mapping relationship between hue and luminance under different lighting conditions.For the region of poor connectivity in the extracted image,the nearest-neighbor skeleton connection method is proposed to enhance the connectivity of the object region,so as to make it constitute a complete object contour.By the detection comparison experiments of multi-scene object extraction,the overall average detection accuracy rate of the proposed method is improved by 0.102 6 in comparison with that of the morphological processing method.Therefore,the object extraction method proposed can improve the object detection accuracy rate of the unmanned detection platforms in complex scenes.关键词
颜色空间/阈值分割/颜色增强/目标提取/骨架提取/最近邻连接算法Key words
color space/threshold segmentation/color enhancement/object extraction/skeleton extraction/nearest-neighbor connection algorithm分类
电子信息工程引用本文复制引用
李明禄,赵春青,侯茂新,王肖霞,杨风暴..面向复杂场景目标提取的颜色增强与连通区域构造方法[J].现代电子技术,2025,48(13):20-28,9.
