Research On Automatic Stitching Technology Of 3D Model Based On Local Feature

Aiming at the virtual assembly module in ”Remote Sensing Satellite Design System”project,we implement an automatic splicing method to solve the engineering problem of automatic splicing of satellite components in the system and to minimize the manual intervention in the splicing process.Among the existing technologies,there are two typical types:Automatic assembly in virtual assembly technology;Automatic stitching of artifacts.However,it does not apply to the application scenario of this paper.Therefore,by combining three-dimensional model retrieval techniques,this paper studies the feature extraction and matching methods between three-dimensional models,and analyzes the feature extraction methods of three-dimensional models to find a local feature matching method suitable for satellite components to determine their splice interfaces and to achieve automatic stitching between the two.In order to solve the above problems,firstly,this paper proposes a method for segmenting a 3D model and extracting features and locally matching for each part of the surface to find the splicing interface between the splicing models,and realizing the automatic splicing between models according to the coincidence between interfaces.;Secondly,a local feature matching method is proposed for 3D model based on loworder surface blending features.Partial surfaces of the segmented 3D model are subjected to feature extraction and similarity matching to find partial surfaces with high similarity between models.Finally,a three-dimensional local feature matching method based on surface sets is proposed to detect the position relationship between surfaces,constrain the combination of surfaces,and calculate the combination matching degree of the surface set,Determine the splicing interface between models from the combination of multiple surface sets and achieve automatic stitching according to the relationship between the interfaces of the model for space transformation.We tested the proposed algorithm and the original algorithm on the rocket model used in the ”Remote Sensing Satellite Design System”,and proved that the proposed method for local feature extraction of 3D models based on hybrid features of low-order surfaces has a higher matching accuracy.Meanwhile,the three-dimensional local feature matching method based on the surface set can also accurately determine the splicing interface between the models and complete the automatic splicing.