Multi-mode dense fog removing method based on visible light-far infrared image

Abstract

The invention discloses a multi-mode dense fog removing method based on a visible light-far infrared image. Comprising the following steps: 1) collecting visible light and far infrared images by using a multi-view camera and an ultrasonic nebulizer, and obtaining a visible light-far infrared image group with the same view field and resolution ratio through image registration as the input of a'multi-modal image'fusion-migration 'collaborative defogging network'; step 2), constructing a'multi-modal image'fusion-migration 'collaborative defogging network'; and step 3) using a multi-modal image data set as training data, and using a'multi-modal image'fusion-migration 'collaborative defogging network' as a generator. Meanwhile, a discriminator is established, an adversarial generative network model is adopted for training, and the optimal network parameters of the multi-modal image fusion-migration collaborative defogging network are obtained; and 4) performing dense fog removal processing on the visible light-far infrared multi-modal image by using the trained multi-modal image fusion-migration collaborative defogging network. The method can effectively fuse the features of the multi-modal image, realize modal complementation, and significantly improve the effectiveness of image defogging under the dense fog condition.

Description

technical field [0001] The present invention relates to an image-based defogging technology, in particular to a fog-oriented scene, using multi-modal images (visible light-infrared image pairs) as input signals, through "multi-modal image 'fusion-migration' cooperative defogging" Fog network" is a design method for multi-effect de-fog. Background technique [0002] With the rapid development of computer vision technology, it has been widely used in scientific research fields such as video surveillance, intelligent transportation, and object tracking. However, these outdoor computer vision researches are extremely vulnerable to weather conditions, and the frequent haze weather in recent years is a very big threat. In hazy weather, due to the small particles or water droplets floating in the air, serious scattering occurs under the sunlight, which makes the pictures captured by the camera equipment seriously degraded. Therefore, in order to ensure that the research of comput...

AI Technical Summary

Problems solved by technology

[0004] In summary, the existing image-based dehazing methods are limited to the research on visible light color images, that is, the content and color information of visible light color images are used to dehaze

Images