Oblique Aerial Image Matching Of Forest 3D Scene

Forest is an important natural resource of our country.To effectively manage forest resources,we need to investigate and monitor them.However,due to the fact that forests mostly grow in mountainous regions where the terrain conditions are more complex,this has caused great inconvenience for manual investigation.The emergence of unmanned aerial vehicle(UAV)photogrammetry technology has made it possible to achieve continuous monitoring of forest resources.This technique can not only rapidly obtain high-resolution multi-angle images in forest areas,but also can acquire three-dimensional modeling of multi-angle data.Through analysis and research on the three-dimensional forest models obtained after modeling,it is possible to extract individual tree data.Parameters such as tree height,diameter at breast height and crown width can be calculated at the same time.The three-dimensional modeling based on oblique images mainly includes the process of multi-view image area network joint adjustment,image fusion,and oblique image matching.Inclination image matching is a key technology in 3D modeling.The matching results directly determine the 3D modeling accuracy.Therefore,it is of great significance to study the matching of oblique images.This paper takes the forest scene oblique aerial images as the research object,by analyzing the current commonly used oblique image matching algorithm,starting from the content of matching primitives,matching strategies,etc.,and develops an oblique aerial image matching method suitable for forest three-dimensional scenes.It greatly improves the accuracy of 3D forest modeling and provides a guarantee for the accurate extraction of subsequent forest parameters.The main research results are as follows:(1)According to the characteristics of forest covered with each other in three-dimensional aerial images of forests,the features of feature matching are used as matching primitives,and a pyramid multi-level matching strategy is used to fully combine the advantages of the two to achieve matching oblique images.(2)Through the comparison and analysis of several feature extraction,feature description and feature matching algorithms commonly used in oblique image matching,combined with the respective advantages of Harris,Harris-Laplace,and Speeded Up Robust Features(SURF)algorithms,this paper put forward the improved method named HLS of Harris-Laplace feature extraction and SURF descriptors based on the combination of the scale invariant feature transform(Scale Invariant Feature Transform SIFT)algorithm in feature matching method matching images,then the feature matching method in Scale Invariant Feature Transform(SIFT)algorithm is used to complete the matching of oblique images,and the results are consistent through random sampling.The Random Sample Consensus(RANSAC)method is used to remove the mismatched pair of points.(3)The feature extraction and feature matching experiments are performed on the forest scene under the visual and oblique images.The Harris operator,the SIFT operator,and the performance of the improved method presented in this paper are compared experimentally.Compared with the other two algorithms,the results show that the method has better stability in scaling,rotation transformation and translation and noise interference;this method is also better than the SIFT algorithm in anti-affine transformation and matching speed,and the proposed method can be applied to the forest three-dimensional scene aerial oblique photography image matching.Therefore,the accuracy of 3D forest modeling can be guaranteed,which provides support for the subsequent fine extraction of forest parameters.