摘要
传统的Canny边缘检测算法容易受到环境噪声和光照变化等外界因素的影响,导致检测结果中会包含部分的背景图中的边缘,而且需要人工选择参数,算法的灵活性和适应性较低。提出了一种改进的基于Canny算子的边缘检测算法,采用双边滤波抑制噪声;利用子图像依据原始背景图均值的方差变化率,确定是否对子图像进行边缘检测与跟踪;使用OTSU算法计算前景亮度自适应阈值及对应的梯度值的阈值以确定Canny边缘检测的单阈值,得到初步的边缘检测结果;最后结合前景的亮度值和梯度值对检测到的结果进行边缘跟踪,得到完整的轮廓。试验结果表明,改进的Canny边缘检测算法不仅具有一定的抗噪性,而且能够利用轧件高温的特点,有效地提取高温物体光辐射造成的弱边缘,抑制背景中因光照、阴影等产生的虚假边缘,是一种检测高温物体有效的边缘检测算法。
The traditional Canny edge detection algorithm is susceptible to external factors such as environmental noise and illumination changes,resulting in detection results that include some of the edges of the background image,and requires manual selection of parameters,with low flexibility and adaptability of the algorithm.An improved edge detection algorithm based on Canny operator is proposed,which uses bilateral filtering to suppress noise;uses the variance change rate of the sub-image based on the mean value of the original background map to determine whether to perform edge detection and tracking on the sub-image;uses the OTSU algorithm to calculate the adaptive threshold of the foreground brightness and the corresponding threshold of the gradient value to determine the single threshold of Canny edge detection,and obtains the preliminary edge detection results.Finally,the detected results are combined with the foreground luminance values and gradient values for edge tracking to obtain the complete contours.The experimental results show that the improved Canny edge detection algorithm not only has a certain degree of noise immunity,but also can take advantage of the high temperature of the rolled parts to effectively extract the weak edges caused by light radiation of high-temperature objects and suppress the false edges generated by light and shadow in the background,which is an effective edge detection algorithm for detecting high-temperature objects.
作者
陈方升
何纯玉
陈亚飞
CHEN Fangsheng;HE Chunyu;CHEN Yafei(State Key Laboratory of Rolling and Automation,Northeastern University,Shenyang 110819,China)
出处
《轧钢》
2021年第5期81-85,共5页
Steel Rolling
基金
国家重点研发计划项目(2017YFB0304103)
中央高校基本科研业务费专项资金项目(N170708020)。
