摘要
为提高皮革抓取铺展过程的准确性,提出一种基于改进Canny算法的皮革铺展图像边缘检测方法。该方法首先在皮革铺展图像分割中采用二维Otsu算法,并引入狼群算法对二维Otsu算法阈值进行优化,以改进传统的Canny边缘检测阈值由人为设定的问题;利用改进后的Canny算法对皮革铺展图像进行边缘检测,从而为后续的皮革抓取铺展奠定基础。结果表明,引入狼群算法优化后的二维Otsu算法,其迭代次数与运行时间都相较于原始Otsu算法有了大幅降低,结构一致性指数提高了0.024,具有更好的图像处理效果;使用改进的Canny算法,相较于原始Canny算法以及Robert、Prewitts等常用边缘检测算法,不仅实现了阈值的自适应设置,得到的皮革边缘线条也更完整清晰,具有更好的抗噪性。将改进Canny算法用于皮革双机械臂抓取铺展,机械臂可准确检测皮革铺展图像的边缘。由此得出,本方法可为后续的铺展抓取奠定基础,具有一定的实用性。
In order to improve the accuracy of leather gripping and spreading process,an improved Canny algo⁃rithm based edge detection method of leather spreading image was proposed.In this method,2D Otsu algorithm was used in leather spreading image segmentation,and Wolf pack algorithm was introduced to optimize the threshold of 2D Otsu algorithm to improve the problem of artificial threshold setting of traditional Canny edge detection.The improved Canny algorithm was used to detect the edge of the leather spread image,which lay a foundation for the subsequent leather grab spread.The results show that compared with the original Otsu algorithm,the iteration times and running time of the 2D Otsu algorithm optimized by Wolf pack algorithm are greatly reduced,and the structural consistency in⁃dex is increased by 0.024,which has better image processing effect.Compared with the original Canny algorithm and common edge detection algorithms such as Robert and Prewitts,the improved Canny algorithm not only realizes the a⁃daptive setting of the threshold value,but also obtains a more complete and clear leather edge line with better antinoise.The improved Canny algorithm was applied to the leather double robotic arms,which could accurately detect the edges of the leather image.It is concluded that this method can lay a foundation for the subsequent spreading and grasping,and has certain practicability.
作者
马海国
MA Haiguo(Xi'an Aeronautical Polytechnic Institute,Xi'an 710089,China)
出处
《中国皮革》
CAS
2024年第8期52-58,共7页
China Leather
基金
陕西省学校科研基金重点项目(KJTD23-01)。
关键词
皮革铺展
边缘检测
OTSU算法
狼群算法
视觉识别
leather spreading
edge detection
Otsu algorithm
Wolf pack algorithm
visual recognition
