Aerial photogrammetry data processing method

Abstract

The invention relates to the technical field of aerial photogrammetry data processing, and concretely relates to an aerial photogrammetry data processing method. The method comprises the following steps: inputting a second training data set into a trained first semantic segmentation network to obtain each corresponding incomplete semantic segmentation image in the second training data set; carrying out contour extraction on each semantic category in each incomplete semantic segmentation image, and then obtaining a contour feature description vector of each optimal contour; inputting each contour feature description vector into a trained classification network to obtain a corresponding classification category, and further obtaining a corresponding predicted semantic region contour; labeling a label of the second training data set by using each predicted semantic region contour; and training a second semantic segmentation network by using the label data to obtain a trained second semantic segmentation network, and performing semantic segmentation on the target aerial photogrammetry data. The problem that the accuracy of a semantic segmentation network is reduced is solved.

Description

technical field [0001] The invention relates to the technical field of aerial photogrammetry data processing, in particular to a method for processing aerial photogrammetry data. Background technique [0002] In the existing semantic segmentation of aerial photogrammetry data, neural network is usually used for processing. However, due to the high cost of labeling the semantic segmentation network, it will consume a lot of human resources; and the aerial photogrammetry data scene obtained in the actual use process is relatively complex, so it is difficult to completely and accurately label the neural network training time manually. The required label data leads to poor semantic segmentation processing effect and low accuracy after training. Contents of the invention [0003] In order to solve the problem in the prior art that the semantic segmentation of aerial photogrammetry data using a single network has a low accuracy rate, the purpose of the present invention is to p...

AI Technical Summary

Problems solved by technology

[0003] In order to solve the problem in the prior art that the semantic segmentation of aerial photogrammetry data using a single network has a low accuracy rate, the purpose of the present invention is to provide a processing method for aerial photogrammetry data, and the adopted technical solution is as follows:

Images