曲线拟合是图像分析中非常重要的描述符号。最常用的曲线拟合方法是最小二乘法,然而一般的最小二乘法有一定的局限性,已经有不少学者对其进行了一些改进。进一步对最小二乘法进行改进,提出一种新的分段直线拟合算法来代替多项式曲线拟合,以达到简化数学模型的建立和减少计算的目的,使其能够更好地对点序列进行拟合。(Curve fitting is a very important descr iptor in image analysis,the most commonly used curve fitting method is least-squares method.But ordinary least-squares method has some limitations,and there are many scholars have made study of improving it.The authors made further improvement on least-squares method and proposed a new piecewise linear fitting algorithm instead of polynomial curve fitting.The new algorithm achieves the goal of simplifying the mathematical model,reducing the calculation,and makes it better to fit point sequence.)