摘要
研究基于灰狼算法的室内三维空间图像分割方法,提升室内三维空间图像分割结构相似度与效率。通过创建平面直角坐标系、纠正与配准楼层位图和采集三维空间要素图层获取室内三维空间矢量化数据,以该数据为基础利用机器人探索空间,获取三维信息,结合最优点到平面的距离,优化室内三维空间图像帧中所有点,建立最优室内三维空间模型,通过灰狼算法的围捕和攻击等狩猎操作优化K均值聚类函数,形成聚类-灰狼算法的图像分割方法,实现室内三维空间图像分割。实验结果显示:该方法采集的图像要素点比实际位置偏移最多不超过2mm;可以准确建立室内三维空间模型;且分割后的室内三维空间图像信息清晰;与灰狼算法优化前相比,峰值信号比和结构相似度分别提高60.5%和15.5%,运行时间缩短了42.1%。
Study the indoor 3Dspace image segmentation method based on Gray Wolf algorithmto improve the structural similarity and efficiency of Indoor 3D space image segmentation.By creating the plane rectangular coordinate system,correcting and registering the floor bitmap,and collecting the three-dimensional space element layer,the indoor three-dimensional space vectorization data is obtained.Based on the data,the robot is used to explore the space and obtain the three-dimensional information.Combined with the distance from the best point to the plane,all points in the indoor three-dimensional space image frame are optimized and the optimal indoor three-dimensional space model is established.The K-means clustering function is optimized by hunting operations such as the capture and attack of the Gray Wolf Algorithmto formthe image segmentation method of the clustering Gray Wolf Algorithmto achieve indoor 3D space image segmentation.The experimental results showthat the maximumdeviation of image element points collected by this method from the actual position is no more than 2mm.It can accurately establish the three-dimensional space model of indoor space;and the segmented indoor three-dimensional space image information is clear.Compared with the Gray Wolf Algorithm before optimization,the peak signal ratio and structural similarity are increased by 60.5% and 15.5% respectively,and the running time is shortened by 42.1%.
作者
何煊强
崔文涛
HE Xuan-qiang;CUI Wen-tao(Minxi Vocational&Technical College,Longyan 364021,China;Dalian Aitefosi Information Technology Co.,Ltd,Dalian 116000,China)
出处
《遵义师范学院学报》
2023年第5期81-85,109,共6页
Journal of Zunyi Normal University
基金
国家教育部“职业教育建筑钢结构工程技术专业教学资源库”项目(2017-B17)。
关键词
矢量化采集
三维空间建模
K均值聚类
灰狼算法
空间图像分割
vectorization acquisition
3D spatial modeling
K-means clustering
Grey Wolf Algorithm
spatial image segmentation
