微光成象系统中实时降噪的优化设计

Optimum Design of a Real Time Noise Reduction of Low Light Level Image System

微光电视工作于极低照度或恶劣环境下，输出信噪比明显降低，目标完全淹没在背景噪声中，使探测率大大下降。提出了用帧间处理技术消除加性噪声。论述了字长和滤波系数的选择原则。研究了实时去噪的几项关键技术、算法和符号位的设计，各部分工作时间分配以及大背景噪声下时钟同步方法。给出了信噪比改善度的理论计算公式，对所建立的系统进行了实测，字长由原来的１０位减少到８位，信噪比改善仍有１２ｄＢ。不使用专用器件，整机硬件量减少到原来的２／５。使用该技术，可显著提高微光成象和红外热成象装置探测目标的能力。

Low light level TV is usually operated under extremely low illuminance or other adverse circumstances and the output SNR decreased to such a level that possibility of detection is greatly reduced.The paper presents the interframe processing for reduction of additional noise and discusses the principle for determination of word length and filter coefficient.Several key techniques including real time noise reduction,design of algorithms and sign digit,time allocation for all working parts,and clock synchronizing with a background of high noise are investigated.The theoretical formula for computation of improvement on SNR is provided.The practical measurement has been carried out for the constructed real time system and the SNR is improved by 12 dB through the word length reduction to 8 digits from 12 digits.Without special device,the system hardware is reduced to two fifth of the original quantity.By this technique,an obvious improvement can be achieved for the detection capability of low light level imaging device and infrared thermal imaging device.