Cashmere and wool optical microscope image recognition method based on generative adversarial network

Abstract

The invention discloses a cashmere and wool optical microscope image recognition method based on a generative adversarial network, and the method comprises the steps: collecting cashmere and wool materials, making the cashmere and wool materials into fiber samples, obtaining optical images of the surfaces of all the fiber samples through an optical microscope, carrying out the extraction of the fiber images, carrying out the depth discrimination, utilizing a clear light-color fiber image to carry out fiber discrimination task training on a classification network based on deep nerves; carrying out style conversion task training on the generative adversarial network based on an attention mechanism, a classification loss function and a cyclic stability loss function by using an unclear deep color or mottle fiber image, and repeating the steps for a new cashmere and wool material to be tested, or directly sending the clear light-color fiber image to the deep neural classification network for type discrimination, or sending the unclear dark-color or mottled fiber image to the deep neural classification network for type discrimination after being subjected to generative adversarial network style conversion, and finally outputting a discrimination result.

Description

technical field [0001] The invention belongs to the field of detection and analysis in the textile industry, in particular to a cashmere wool optical microscope image recognition method based on a generative confrontation network, which is used to analyze the quality components of special animal fibers after different processes. Background technique [0002] Nowadays, there are many detection technologies and methods for cashmere wool fiber identification, such as fiber optic image recognition method, DNA method, liquid protein analysis method and so on. But so far, these methods cannot be widely used due to their own defects. Therefore, the traditional optical microscopy method is still relatively reliable, mature and accurate identification method. [0003] The traditional optical microscope method is to place a certain amount of plush fibers under a magnified microscope after opening, and to identify them one by one by manually observing the structural features such as s...

AI Technical Summary

Problems solved by technology

However, this method is time-consuming and labor-intensive, thus deriving a technical demand for artificially generating equivalent data using limited data

Images