摘要
为解决不同环境下的工业零件的角点检测的问题,在研究不同种类特征点检测各存在速度与检测精度的差异的基础上,提出一种基于图像角点灰度特征信息与曲率曲度相结合的特征点检测算法。通过结合角点灰度特征分析与类曲度检测特征点的优势,提高相应特征点的检测定位精度与检测成功率,同时兼顾检测时间效率。实验结果表明,与各自原算法相比,Harris-CPDA算法在相同时间内,可提供更为精准的特征点,在实际应用场景中对特征点检测的精度提升具有较大的作用,实现传统检测与深度学习的图像检测的新融合。
To solve the problem of angular point detection of industrial parts under different environments,on the basis of studying the difference between the speed and accuracy of different kinds of characteristic point detection,a characteristic point detection algorithm based on the combination of image angle point grayscale characteristic information and curvature was proposed.By combining the advantages of angle grayscale characteristic analysis with class curvature detection feature points,the detection and positioning accuracy and detection success rate of the corresponding feature points were improved,while taking the detection time efficiency into account.The Harris-CPDA algorithm can provide more accurate feature points in the same time,which has greater effects on the accuracy improvement of feature point detection in actual application scenarios,and realizes a new fusion of traditional detection and deep learning image detection.
作者
胡晓彤
朱博文
程晨
HU Xiao-tong;ZHU Bo-wen;CHENG Chen(College of Artificial Intelligence,Tianjin University of Science and Technology,Tianjin 300457,China)
出处
《计算机工程与设计》
北大核心
2021年第2期504-511,共8页
Computer Engineering and Design
关键词
哈里斯
曲率尺度空间
角点
工业零件
深度学习
Harris
curvature scale space
angular point
industrial parts
deep learning
