图像的退化不变量研究

Study of image degradation invariant

图像的不变量通常指几何不变量,是提高计算机视觉系统自适应性的重要理论。但图像也经常发生灰度的变换或者退化(如,噪声、光照变化、运动模糊等),单纯的几何不变量不能满足实际中目标识别的要求。灰度退化和几何变换都看作图像退化,提出了图像退化不变量的概念,分析了退化不变量的构造,并分析经典的几何不变量的退化不变性。仿真实验结果表明:只有尺度不变量特征变换(SIFT)是较好的退化不变量,可以直接用于未知退化状况的图像匹配。

Image invariant is generally metaling the geometric invariant,which is an important theory to improve the adaptability of computer vision system. But image gray is often degraded, such as noise, illumination change and motion blur. Geometric invariant can not fit the requirements of object recognition in practice. Image gray degradation and geometric transform are treated as image degradation, and image degradation invariant is proposed ,and the construction of degradation invariant is analyzed, The degradation invariance of several classical geometric invariants is analyzed. The simulation experiments show that SIFT is a good degradation invariant and can be directly applied to image matching of unknown degradation.