Terahertz image nondestructive testing method, device and equipment

Abstract

The invention discloses a terahertz image nondestructive detection method, device and equipment and a computer readable storage medium, and the method comprises the steps: inputting a to-be-detected terahertz image into a deep convolutional neural network, and extracting a feature map; converting the feature map into an attention map by using the attention mechanism, the data attribute of the terahertz image, a first target relation matrix and a first target coding matrix; converting the attention graph into a target image feature by using an attention mechanism, the data attribute, a second target relation matrix and a second target coding matrix; converting the target image features into a standard common feature; and determining a damage classification result of the terahertz image by utilizing the standard common feature, the target classification weight matrix and the Softmax function, thereby determining the damage type and damage degree of the terahertz image. According to the method, the device, the equipment and the computer readable storage medium provided by the invention, the damage characteristics of the terahertz image are accurately represented by utilizing multipleattention mechanisms.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a method, device, equipment and computer-readable storage medium for non-destructive testing of terahertz images. Background technique [0002] In recent years, scholars at home and abroad have used terahertz imaging technology to conduct related research on non-destructive flaw detection of materials, especially in the later image enhancement of flaw detection imaging. At present, the technical methods of non-destructive testing of terahertz images mainly include: non-destructive testing of terahertz images based on wavelet transform and Lipsitz exponent and non-destructive testing methods based on wavelet denoising. The non-destructive detection of terahertz images based on wavelet transform and Lipsyntz index loses damage attribute information, and the representation of damage features is very vague, and the final classification results need to be realized man...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a method, device, equipment and computer-readable storage medium for non-destructive testing of terahertz images, so as to solve the problem that the non-destructive testing methods provided by the prior art cannot accurately represent the damage characteristics of terahertz images

Images