小波投影算子在工业图像边缘检测及抗噪分析中的应用

采用基于小波变换投影算子的快速多尺度边缘检测迭代算法，计算表明，迭代次数取5－9，输出图像信噪比较佳；并与传统的Marr边缘检测方法作了实验比较分析，小波变换方法针对工业现场采集的监视目标图像污染严重及噪声干扰等实际情况是一种较好的方法，在边缘定位精度，图像细节丢失等方面优于Marr方法，对噪声干扰也有较好的鲁棒性。

AN APPLICATION OF WAVELET PROJECTIVE OPERATOR TO INDUSTRIAL IMAGE EDGE DETECTION AND ANALYSIS OF ANTI- NOISE

As the matter stands, images of detected objects in industrial workshop are always seriously blurred and disturbed by noise. In this paper, a fast algorithm used in multiscales edge detection based on projective operator of wavelet transform is presented. When steps of the algorithm is chosen to be 5-9 steps, the result shows that the signal/noise ratio (SNR) of output image is better. We compare the fast algorithm with the traditional Marr edge detection method. Analysis and experiment result show that wavelet transform method is better, especially in precision of edge location and lose of image details. Wavelet transform method has also better robustness to niose.