Ship sample expansion method in infrared image based on style generative adversarial network

Abstract

The invention provides a ship sample expansion method in an infrared image based on a style generative adversarial network. The method mainly solves the problems that in the prior art, the authenticity of generated infrared images is poor due to complex simulation modeling, the collection difficulty of visible light-infrared image photoelectric conversion training samples is large, and the numberof training sets is small, so that the expanded infrared images are lack of diversity. The method comprises the following steps: (1) selecting real-shot infrared images to form the training sets; (2)constructing a generator network; (3) constructing a discriminator network; (4) constructing a style generative adversarial network; (5) training a discriminator network; (6) training a generator network; (7) training a style generative adversarial network; and (8) outputting an infrared image sample by using the trained generator network to complete the expansion of the infrared image sample. According to the method, a large number of infrared ship samples can be generated, and the sense of reality and diversity of expanded samples are effectively improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a method for expanding ship samples in infrared images based on styleGAN (style-based Generative Adversarial Network) in the field of deep learning. The invention can expand the ship samples in the infrared image so as to provide abundant data sets for the training of the infrared target detection and recognition algorithm. Background technique [0002] Infrared imaging technology is often used in target detection, identification and tracking due to its strong target detection ability and strong anti-interference ability. Due to the complex infrared characteristics of the target, which change significantly with temperature conditions, it is difficult to detect and recognize infrared targets. In order to improve the ability to detect and recognize infrared targets, a large number of infrared images are generally required to train the detection and recognition algori...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art, and propose a method for expanding ship samples in infrared images based on pattern generation confrontation network, aiming at solving the problem of complex simulation process when expanding infrared image samples and the realism of extended infrared samples. Poor, the photoelectric conversion method from visible light to infrared is difficult to obtain the training set, the expansion of infrared samples lacks diversity, and the number of expansions is limited

Images