摘要
当前的动态目标识别方法在场景复杂的图像中,因为无法采集足够多的特征信息,导致识别结果应用性受限。基于三维激光点云提出一种运动图像动态目标识别方法。利用三维扫描系统获取运动点云图像特征,在不影响有效信息采集的情况下,进行图像预处理;引入地平面方程,将图像背景点云与被识别目标点云通过欧式聚类法分割,提取处理后的被识别目标关键点,并采用Freeman链码检测边缘特征,完成运动图像动态目标识别。试验对比结果表明,所研究基于三维激光点云的运动图像动态目标识别方法,对动态目标有良好的鉴别能力及较好的识别精度,且所需动态目标识别时间较短。
The current dynamic target recognition methods cannot collect enough feature information in complex images, which leads to the limitation of the application of recognition results. Based on 3 D laser point cloud, a moving image dynamic target recognition method is proposed. The 3 D scanning system is used to obtain the image features of point cloud, and the image preprocessing is carried out without affecting the effective information acquisition. The ground plane equation is introduced, and the image background point cloud and the identified target point cloud are segmented by Euclidean clustering method. The key points of the identified target are extracted after processing, and the Freeman chain code is used to detect the edge features, so as to complete the dynamic target recognition of the moving image. Experimental results show that the dynamic target recognition method based on three-dimensional laser point cloud has good identification ability, good recognition accuracy and short dynamic target recognition time.
作者
尤锟
张全成
李凯勇
You Kun;Zhang Quancheng;Li Kaiyong(Xi’an Shiyou University,Xi'an Shaanxi 710065,China;Physics and Electronic Information Engineering,Qinghai Nationalities University,Xining,Qinghai 810007,China)
出处
《应用激光》
CSCD
北大核心
2022年第4期147-153,共7页
Applied Laser
基金
陕西省自然科学基础研究计划(2020JQ-736,2019JM-154)。
关键词
动态目标识别
三维扫描系统
点云图像
图像预处理
关键点提取
边缘特征检测
dynamic target recognition
three-dimensional scanning system
point cloud image
image preprocessing
key point extraction
edge feature detection
