基于改进卡尔曼滤波在Canny边缘检测中的应用

Canny Edge Detection Algorithm Based on Improved Kalman Filter

在视觉检测中,为了能够精确地辨识出图像的边缘,在现行算法的基础上设计了一种基于新的滤波方法的Canny边缘检测算法。该滤波算法采用Kalman空间模型代替原高斯函数来平滑图像,以最小均方误差为最佳估计准则,优化滤波后图像质量。并针对滤波过程中存在的漏检误检以及信噪比问题,将分数阶微分掩模应用于最优递推方法中,从而结合了两者的优点,达到相互优化效果。实验表明,该方法对对比度较低的二维数字图像都有很好的效果,在增强图像边缘细节的同时,有效降低了图像中的噪声。

An improved Canny edge detection algorithm is proposed on a basis of studying current various edge detection algorithms, in order to recognize the edge precisely in the visual measuring. The Kalman spatial model is used to instead of original Gaussian function to filter the object,with the quality of filtered image optimized with the minimum MSE as the best estimation criterion. In order to decrease the detector errors and to enhance SNR, the fractional calculus module is applied to the Kalman optimal recursive algorithm. Advantages of both two are integrated. The simulation results indicate that the algorithm is efficient in processing the low - contrast image and weakening the noises without blurring the edge details.