3D Reconstruction And Morphological Analysis Of Butterfly Based On Monocular Vision

The bionic flapping-wing micro aerial vehicle has a high research value in both civil and military fields because of the advantages of good flexibility and high stealthiness.Due to the complicated flight process of butterfly,the research on butterfly flight mechanism has not yet formed a complete theory,and the research on the bionic butterfly micro aerial vehicle is also relatively less.The shape profile and motion attitude of the butterfly are important parameters for the design of the bionic micro aerial vehicle,so the research on the morphological analysis of the butterfly is very necessary.The current research methods require high accuracy of measurement instruments,long experimental period,low efficiency,and are not universal.To address the above problems,this paper proposes a method of butterfly 3D reconstruction and morphological analysis based on monocular vision,which provides reference for the design of the bionic micro aerial vehicle.The main research contents of this paper are as follows:To address the problem that background noise can interfere with butterfly 3D reconstruction and thus lead to low accuracy,a butterfly target extraction algorithm based on mathematical morphology and threshold segmentation is proposed.Firstly,the video frame image is pre-processed,mainly including video frame splitting,image grayscale conversion and image filtering.Secondly,the gradient of the image is solved by Sobel operator,followed by the maximum entropy method to obtain the binary image by thresholding the gradient image,so as to extract the preliminary outline of the butterfly.Then,in order to obtain the complete butterfly region,morphological operations such as expansion,erosion and hole filling are performed on the binary image first,followed by removing irrelevant connected components and performing image restoration operations to achieve effective extraction of the butterfly target.Finally,experimental validation and comparative analysis of the butterfly target extraction algorithm are conducted,and the results show that the algorithm proposed in this paper has good feasibility and flexibility.For the problem of unknown camera internal parameters in butterfly 3D reconstruction,a self-calibration method of camera based on genetic algorithm is proposed.First,the ORB algorithm is used to obtain the feature points in the image,and the dynamic feature points are filtered out by combining the extraction results of butterfly targets.Secondly,the K-nearest neighbor algorithm is used for feature point matching.Then,the basis matrix is solved according to the matching relationship of feature points between images,and the accuracy analysis of the basis matrix is performed to obtain the basis matrix with high accuracy.Finally,the singular value decomposition of the basis matrix is performed to derive the simplified Kruppa equation,and the objective function established based on the Kruppa equation is optimally solved using genetic algorithm to obtain the camera internal parameters.For the problem of reconstructing and morphological analysis of dynamic butterfly targets,a method is proposed to construct a three-dimensional model of butterfly based on image optical flow.Firstly,the Lucas-Kanade optical flow method is used to solve the motion velocity of all pixel points of the butterfly.Secondly,a 3D motion-by-point linear estimation model is constructed based on the relationship between the 2D image optical flow field and the 3D scene motion field,so as to solve the relative 3D coordinates of all pixel points of the butterfly and realize the 3D reconstruction of the point cloud.Then,the 3D model of the butterfly is contoured to obtain the structural parameters such as wingspan,heel chord length and wing chord length.Finally,the motion attitude of the butterfly is analyzed according to the motion velocity of the target feature points of the butterfly,and the flight parameters such as flapping angle are solved.