基于迭代拟合的集装箱角件边缘检测算法

噪声通常是影响集装箱角件图像中低层次语义信息提取精度的重要因素,传统的边缘检测算法通常通过改进滤波器和阈值来消除图像中的物理噪声和环境噪声,但是却无法去除边缘检测后的噪声,为解决这一问题,提出了一种基于迭代拟合的边缘检测算法。首先,对角件图像进行一系列预处理操作获取边缘点集,其次,使用拟合算法处理点集并且得到函数表达式,然后定义偏差值度量并计算,用于衡量目标点集到拟合或者检测结果的偏差,最后,去除定义下距离拟合结果最远的指定数量的点,如此迭代拟合直至评价函数收敛。实验结果与分析表明,该算法可以有效地去除边缘点集中的非真实边缘点,相比于传统的边缘检测算法更能去除特殊噪声,算法具有收敛速度快、准确率较高、灵活性好等特点。

Edge Detection Algorithm for Container Corner Pieces Based on Iterative Fitting

Noise is usually the important factor that affects the accuracy of low level semantic information extraction in container corner pieces images. Traditional edge detection algorithm usually reduces the physical noise and ambient noise in the image by improving the filter and the threshold, but cannot remove the noise after edge detection. To solve the problem, an algorithm of edge detection based on iterative fitting was proposed. Firstly, container corner pieces image was processed by a series of preprocessing operations to obtain the set of edge points. Secondly, the fitting algorithm was used to process the set of points and the function expression was obtained. Thirdly, the deviation value was defined and calculated to measure the deviation of the target point set to the fitting or detection result. Finally, iteratively removing some points of farthest de-fined distant to fitting result until evaluation functions converge. The result of experiments shows that the proposed algorithm can effectively remove the unreal edge points and better than traditional edge detection algorithm in removing special noise. The proposed algorithm several merits, such as fast convergence rate, higher accuracy rate and good flexibility