基于低秩矩阵恢复的高噪火场影像去噪与增强技术研究

Research on Denoising and Enhancement Technology for High-Noise Fire Scene Images Based upon Low-Rank Matrix Recovery

火灾影像处理与分析技术在火灾监测、火灾事故救援以及火灾调查等方面发挥着至关重要的作用,而火灾影像分析的精准度与待处理火灾影像的质量关系密切。尤其在高噪声污染背景下如何做好火灾影像质量的提升工作更为重要。在图像处理中,低秩矩阵恢复技术能够挖掘图像本身的低秩特性,在一定程度上降低噪声强度的影响,呈现了优异的去噪特性。基于此,本文针对火灾事故实境案例火灾视频影像帧数据,叠加高噪声影响因子,利用图像低秩性构建低秩矩阵先验模型,通过算法迭代得到低秩矩阵逼近,从而建立高噪声火灾影像的恢复模型,与传统中值滤波去噪方法相比,去除噪声点效果更佳,能够更好地保留图像边缘信息,具有良好的高噪声火灾影像处理精度提升效果,从而助力提高火灾监测、指挥决策与应急救援能力水平。

Fire image processing and analysis technology plays a vital role in fire monitoring,fire accident rescue,and fire investigation.The precision of fire image analysis is closely related to the quality of the fire image to be processed,especially in the context of high noise pollution background,it is particularly important to improve the fire image quality.During image processing,the low-rank matrix recovery technology can explore its low-rank characteristics by itself,reducing the impact of noise intensity to some extent and showing excellent denoising characteristics.According to this,the paper regarding the video frame data of real-life fire video images in real cases,superimposes high-noise impact factors,and constructs a prior model of low-rank matrix by using the low-rank property of images.Through algorithm iteration,the low rank matrix approximation is obtained to establish a restoration model for high noise fire images.Comparing with traditional median filtering denoising method,the denoising effect of the low-rank matrix recovery method is better,which can better preserve image edge information and has a good effect on improving the accuracy of high noise fire image processing,thereby helping to improve the level of fire monitoring,command decision-making,and emergency rescue capabilities.