Multi-exposure infrared image fusion method

Abstract

The invention discloses a multi-exposure infrared image fusion method. The multi-exposure infrared image fusion method comprises the following steps: obtaining a contrast array graph, a noise point invisibility array graph and an exposure suitability array graph of a plurality of to-be-fused infrared images; obtaining a normalized weight map of each infrared image according to the contrast ratio array map, the noise point non-visibility array map and the exposure suitability array map; performing Gaussian pyramid decomposition on the normalized weight map to obtain a weight map Gaussian pyramid of each infrared image; performing Laplacian pyramid transformation on each infrared image to obtain a Laplacian pyramid of each infrared image; and carrying out fusion on the Laplacian pyramids ofthe plurality of infrared images by using the weight map Gaussian pyramid to form a fused infrared image. According to the fusion method, detail expression information of the infrared image is enhanced, and the influence of infrared image noise points on the image quality can be inhibited.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a multi-exposure infrared image fusion method. Background technique [0002] Infrared images are a representation of scene temperature differences. Generally, the greater the scene temperature difference, the stronger the layering of the image, and the easier the information contained is to distinguish. Most of the infrared source data collected by the thermal imager is 14 bits. If the dynamic range of the scene temperature difference exceeds the dynamic range that the thermal imager can represent, the infrared image formed by the thermal imager will have over-exposed and under-exposed areas. Such over-exposure and under-exposure areas affect people's discrimination of thermal information of ultra-high temperature and ultra-low temperature target details, and at the same time reduce the visualization effect of infrared images. Therefore, multi-exposure fusion...

AI Technical Summary

Problems solved by technology

This method adds a Gaussian weight function to smooth the weighted image in order to make the connection effect look natural, so that there is no obvious brightness mutation area at the edge of the image block, but in the area where the brightness change is relatively strong, there is still brightness connection between different image blocks The problem of unnaturalness, if this method is used for fusion of infrared images, the edge of the fusion effect will be unnatural, and the area with large noise in the infrared image will still be introduced as the fusion object, which will reduce the fusion quality of the infrared image

Images