Adaptive subregional recombination method for high-and-low-frequency signals in infrared image data processing

Abstract

The invention relates to an adaptive subregional recombination method for high-and-low-frequency signals in infrared image data processing. According to data characteristics of high-and-low-frequencysignals in infrared image data processing, subregional extraction of image feature factors is carried out on the infrared low-frequency signals; a mathematical model capable of controlling a high-and-low-frequency signal recombination proportion adaptively based on the image feature factors is established; and then the high-and-low-frequency signals are recombined based on the mathematical model to obtain an infrared image with a high frequency part with clear details and a low-frequency part with a good noise suppression effect. According to the invention, engineering implementation of the method is realized; detail information of the original image is improved and noise amplification in a flat region is suppressed, so that the recombined image contains abundant detail information withoutamplified background noises; and thus the recognition efficiency of the specific scene target is increased obviously.

Description

technical field [0001] The invention belongs to the field of image processing, and relates to an adaptive sub-area recombination method for high and low frequency signals in infrared image data processing. Background technique [0002] Due to the characteristics of high dynamic range of infrared scene information, thermal imaging systems usually use A / D conversion of more than 14bit for the output signal of the detector, but most of the video output The output signal is converted to an 8bit video output signal through compression mapping methods such as linear transformation (AGC) or histogram equalization (PE). Due to the dynamic compression methods such as AGC and PE, although simple and effective, there are still problems such as over-enhancement, noise amplification in flat areas, and bleaching effects. In order to solve these problems, image detail enhancement is needed to reduce detail loss. However, the infrared image processed by the original image detail enhanceme...

AI Technical Summary

Problems solved by technology

However, the infrared image processed by the original image detail enhancement algorithm has the problem of over-enhancement of the noise signal in low-contrast scenes, so that in low-contrast scenes, such as sea-sky scenes and open-ground scenes, using the original image detail enhancement algorithm in the sky area there will be a lot of noise

[0003] In order to solve the above problems, Jin Weiqi et al. of Beijing Institute of Technology proposed a method to directly control the reconstruction ratio of high-frequency images according to the image grayscale or texture. This method directly controls the reconstruction ratio by grayscale or texture information, although it can be used in pure noise The area can suppress the enhancement factor of the noise, but in the critical area of ​​noise and details, due to the slow convergence, the reconstructed image will have a large amount of noise residue at the edge of the image

The multi-layer image adaptive amplification technology proposed by Qian Jia et al. of Nanjing University of Science and Technology divides high-frequency details into multiple layers and controls the proportion of multi-layer details to increase. However, due to the exponential growth of the scale parameter, the high-frequency details The enhancement is not continuous, resulting in a harsh visual effect of the restructured image

Images