科学观测的目的是获知对象的真实面貌，由于物理原理或技术条件的限制以及噪声的介入，我们常常只能获得一个降质的信号（包括图像）。信息传递、变换和应用的过程可能伴随信号的降质。信号复原是一类技术方法，它依据降质的观测来估计原来不失真的信号和图像。该技术的主要内容是反卷积；信号和图像估计；噪声抑制。反卷积和信号复原是信号处理技术中具有理论挑战性的分支。由于应用广泛，它一直是研究热点。本文强调了反卷积问题的物理起源、理论方法的要点、适应范围和限制； 由于水平和时间限制，内容仅限于信号处理、线性代数和随机过程的基础知识，反卷积的基本理论，一维信号反卷积的Levinson-Durbin算法和实现。

The purpose is to learn scientific observations the true face of the object, as physics or technology constraints, and the involvement of noise, we often only get a degraded signal (including images). Information transmission, transformation and application process may be associated with lower quality signal. Signal recovery is a kind of technical approach, which lower the quality according to the original observations to estimate the signal and image without distortion. The main contents of the technology Deconvolution; signal and image estimation; noise suppression. Deconvolution and Signal Recovery is a signal processing technology branch of theoretical challenges. As widely used, it has been a research focus. This article emphasizes the physical origin of the problem of deconvolution, the main points of theoretical methods, scope and limits of adaptation; as the level and time constraints, the content is limited to signal processing, linear algebra and basic knowledge of stochastic processes, deconvolution of the basic theory, one-dimensional signal deconvolution of the Levinson-Durbin algorithm and implementation.