Research And Realization On Layout Reconstruction Technology Of Indoor Scene Based On Structure Analysis

In the real world,indoor scenes are the most closely related places to people’s daily work and life,and are the most important components of the human perception of the world.With the rapid development of robot technology,computer vision and artificial intelligence,the reconstruction of indoor scene is becoming more urgent-However,due to the diversity of object in indoor scenes,the complexity of the geometric information of objects,the density of objects in the scene,the lack of data caused by the occlusion between objects,and the sparseness of the scene data,the difficulty in automatic reconstruction of indoor scenes has undoubtedly increased,so the research of indoor scene reconstruction is still a challenging task.Therefore,this paper takes the indoor scene as input,and structural analysis is conducted,as follows:(1)A rough division of room layout based on data projection and line detection is proposed.Firstly,the floor and ceiling of the scene data is removed,and then the remained data are projected into the two-dimensional plane.Then the candidate wall data is obtained by detecting the lines after projection through the improved Ransac algorithm,and the improved Mean shift method is adopted to merge the straight line segments belonging to the same wall and thus acquire the effective representative lines.Finally,the preliminary division of the room layout is realized by detecting the lines and its intersection.(2)A room layout subdivision method based on undirected graph is designed.Based on the preliminary division of the room layout,our method labels room elements binary statuses as empty or non-empty according to the ray method.The room preliminary division result is converted into the undirected graph.According to this,the energy equation is established,and the characteristics of the point coverage and the Euclidean distance between the expected room elements are considered to set the bound term to regulate the segmentation process.Eventually the graph-cut algorithm is adopted to implement the room layout results of the indoor scene.(3)A progressive structure reconstruction method of interior scene is implemented.The method extracts the walls based on region growing,and calculates the feature information of the structure wall based on rasterization to analyze the point cloud unit coverage,so as to classify the walls.The single walls can be availably fitted by Oriented Bounding Box.The windows,doors and other openings are detected,cording to the point cloud unit coverage.Different types of walls are integrated to complete the reconstruction of the overall walls.Finally,combined with floor and ceiling,the reconstruction of the scene room structure layout of the indoor scene is achieved.