Image recognition-based automatic detection and analysis method for residential outer wall cracks

Abstract

The present application relates to a kind of based on image recognition's residential outer wall crack automatic detection and analysis method, belong to building structure detection and image recognition cross technical field.For the technical problem that traditional residential outer wall crack detection relies on artificial, high-altitude operation risk is high, efficiency is low and precision is insufficient, its technical scheme main points are as follows: control unmanned aerial vehicle with camera to gather the same outer wall area image with double pitch angle, get difference image by registration difference;Three images are input into pre-training multi-branch convolutional neural network to extract and fuse features, and output crack pixel binary segmentation map;The pixel width is calculated by morphological skeletonization, combined with the positioning ranging data of unmanned aerial vehicle to convert actual physical width, finally generate crack detection report.It is mainly used for the automatic, high-precision detection and analysis of residential outer wall crack, and provides reliable data support for building structure safety evaluation.

Images