Research on dark channel dehazing of single-image based on non-dispersive infrared (NDIR) detection technology

IF 1.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

International Journal of Fuzzy Logic and Intelligent Systems Pub Date : 2021-05-31 DOI:10.3233/JIFS-219089

Xiang-xiang Zheng, Z. Xu

引用次数: 1

Abstract

This paper presents an experimental study on the non-dispersive infrared (NDIR) detection technology and dark channel dehazing technology. Based on the analysis of Beer-Lambert Law and differential carbon dioxide detection principle, this paper proposes an atmospheric light value estimation algorithm based on NDIR detection technology. First, the change characteristics of the gas concentration in indoor smoky environment are collected and analyzed. Then appropriate weighting coefficients are chosen based on the gas characteristics to estimate the atmospheric light value. Finally, the digital image dehazing technology through dark channel prior is used for calculation to obtain a haze-free image with high quality and high resolution. The experiment in this paper proves the feasibility of combining NDIR detection technology with dehazing technology, and its ability to improve image quality and achieve better restoration effect.

查看原文本刊更多论文

基于非色散红外(NDIR)检测技术的单幅图像暗通道去雾研究

本文对非色散红外(NDIR)探测技术和暗通道除雾技术进行了实验研究。本文在分析比尔-朗伯定律和差分二氧化碳检测原理的基础上，提出了一种基于NDIR检测技术的大气光值估计算法。首先，收集并分析了室内烟雾环境中气体浓度的变化特征。然后根据气体特性选择适当的加权系数来估算大气光值。最后，利用暗通道先验的数字图像去雾技术进行计算，得到高质量、高分辨率的无雾图像。本文的实验证明了NDIR检测技术与去雾技术相结合的可行性，能够提高图像质量，达到更好的恢复效果。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

International Journal of Fuzzy Logic and Intelligent Systems COMPUTER SCIENCE, THEORY & METHODS-

CiteScore

2.80

自引率

23.10%

发文量

期刊介绍： The International Journal of Fuzzy Logic and Intelligent Systems (pISSN 1598-2645, eISSN 2093-744X) is published quarterly by the Korean Institute of Intelligent Systems. The official title of the journal is International Journal of Fuzzy Logic and Intelligent Systems and the abbreviated title is Int. J. Fuzzy Log. Intell. Syst. Some, or all, of the articles in the journal are indexed in SCOPUS, Korea Citation Index (KCI), DOI/CrossrRef, DBLP, and Google Scholar. The journal was launched in 2001 and dedicated to the dissemination of well-defined theoretical and empirical studies results that have a potential impact on the realization of intelligent systems based on fuzzy logic and intelligent systems theory. Specific topics include, but are not limited to: a) computational intelligence techniques including fuzzy logic systems, neural networks and evolutionary computation; b) intelligent control, instrumentation and robotics; c) adaptive signal and multimedia processing; d) intelligent information processing including pattern recognition and information processing; e) machine learning and smart systems including data mining and intelligent service practices; f) fuzzy theory and its applications.