Lightweight image fusion algorithm and device based on weak computing power

Abstract

The invention designs a lightweight image fusion algorithm and device based on weak computing power. The lightweight image fusion algorithm includes the steps: obtaining and preprocessing an infraredimage and a visible light image under the same scene at the same moment by using a sensor; using an image fusion algorithm based on total variation minimization, wherein the fusion algorithm is deployed on an S835 development board, and the processed image is used as an H channel; and combining the H channel with a UV channel of the visible light image, to obtain a final result. The lightweight image fusion algorithm optimizes a fusion algorithm for the development board, including the steps: optimizing a solution formula of total variation minimization, optimizing adaptive gamma correction, performing multi-core advantage parallelization processing on the development board, and disassembling internal circulation to enable the number of circulation layers to be not more than two as much aspossible.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a lightweight image fusion algorithm and device based on weak computing power. Background technique [0002] Image fusion refers to combining two or more images of different types into a new image using a specific algorithm to achieve a more complete and clear description of the scene. Image fusion has important application value in the fields of medicine, remote sensing detection, weather forecast, disaster detection and analysis, military target detection, etc., especially in computer vision. [0003] The simple combined image fusion method needs to manually set the weighting coefficients, and different scenes require different weighting coefficients, so the applicability is poor. The logical filter image fusion method needs to set logical operation rules, and different processing needs to be performed for different scenes, so the applicability is poor. The pyramid image fusio...

AI Technical Summary

Problems solved by technology

[0003] The simple combined image fusion method needs to manually set the weighting coefficients, and different scenes require different weighting coefficients, so the applicability is poor

The logical filter image fusion method needs to set logical operation rules, and different processing needs to be performed for different scenes, so the applicability is poor

The pyramid image fusion method needs to build an image pyramid, so the amount of data will be enlarged, and there is redundancy between different layers of the image pyramid, which will increase the computational burden; the pyramid fusion method needs to inversely reconstruct the fusion pyramid, and this process is unstable sex

The wavelet transform fusion method needs to select a suitable wavelet base and set the number of wavelet decomposition layers. Too few decomposition layers will cause some details to disappear after fusion, while too many decomposition layers will consume a lot of time.

These methods have some problems in terms of adaptability or calculation, so a lightweight image fusion algorithm based on weak computing power is proposed.

Images