基于小波变换的红外探测系统信号去噪

Signal Denoising of Infrared Detection System Based on Wavelet Transform

针对红外探测系统目标识别时漏报警与误触发的问题,基于小波分析理论,对转动探测系统输出的红外信号进行小波去噪处理。构造适用于本系统红外信号去噪处理的阈值函数,通过计算分析与实验验证,采用4层coif小波基函数分解效果最佳。本文构造的阈值函数较传统软、硬阈值函数信噪比(SNR)提高17.52%~34.5%,均方误差(MSE)减小16.15%~20.77%,在不丢失原始波形信息的前提下,使无车辆目标时输出波形平坦、有目标时输出波形光滑,为后期实现车辆目标的识别提供理论依据。

This study aims to solve the fault alarm problem in recognition of vehicle targets using infrared detection systems based on the theory of wavelet analysis. The wavelets are used to denoise the output infrared signal from rotation detection system. Therefore, it is important to create a new threshold function suitable for denoising infrared signals in this system. The computational analysis and experimental verification reveal that the four-layer “coif” offers the best wavelet decomposition. Here, the SNR increases by 17.52%-34.5% and the MSE decreases by 16.15%-20.77% compared to the traditional soft ＆ hard threshold function. Moreover, the output waveforms with and without the vehicle target are smooth and flat, respectively. It provides a theoretical basis for accomplishing vehicle target recognition later.