利用噪声消除量化误差的方法

A Method of Utilizing Noise to Remove The Quantization Error

本文提出一种提高图象处理精度的方法.利用自然界中存在的噪声或者人力地介入一个噪声源.使其标准差近于量化误差,对相对稳定信号,作m帧叠加的图象增强,可以提高原图象的量化精度m^(1/2)倍。该方法已由计算机模拟和太阳磁场望运镜相应系统的实测获得证实.通过计算,找到了噪声大小与量化精度间的关系,并确定了该方法的适应范围.根据这一方法,可用商用的相对廉价的,较低精度的A/D变换和图象处理系统,通过图象增强,能获得高精度的结果.

A method to raise the accuracy of digital image processing has been proposed. By means of statistic noise exisied in nature or made artificially, which has the standard deviation approximatele equal to the unit of quantization error, the original quantization accuracy could be raised about m^(1/m) times by imege enhancement of m frames.This method had been confirmed by the computer simulated test and the practical measurements of solar magnetic field. From the computation, the relation between the amount of noise and the quantization accuracy has been found and the applicable arrange of this method has been determined. By meams of this method, the image processing system with A/D converter of low accuracy, which is commercially available and is relatively cheap, could be used to get high accuracy measurement.