| 基于Hu矩和SIFT特征的法兰识别 |
 点此下载全文 |
| 引用本文：朱伟，张美琪，李磊，徐茜茜.基于Hu矩和SIFT特征的法兰识别[J].计算技术与自动化,2026,(1):97-102 |
| 摘要点击次数: 279 |
| 全文下载次数: 229 |
|
|
| 中文摘要:针对二孔法兰识别中Hu矩匹配法兰旋转不变性差、识别精度低的缺点，提出了改进的Hu矩-SIFT匹配。通过对图像进行灰度化和Gamma变换使图像增强，消除金属工件的反光带来的影响，再用混合滤波对图像进行去噪处理，最后对图像进行识别。即利用改进的Hu矩对外层轮廓进行粗匹配，再提取法兰孔洞，对孔洞进行局部SIFT特征点匹配，使用暴力匹配和随机一致性减少误匹配点，由此识别出目标法兰。实验结果表明:改进的Hu矩相比于原来的Hu矩匹配，可以很好地提高匹配精度。 |
| 中文关键词:二孔法兰识别 SIFT特征匹配 混合滤波 Hu矩匹配 |
| |
| Flange Recognition Based on Hu Moments and Sift Features |
|
|
| Abstract:In order to address the shortcomings of poor rotation invariance and low recognition accuracy of Hu moment matching flanges in two hole flange recognition, an improved Hu moment and SIFT matching is proposed. By applying grayscale and gamma transformation to the image for image enhancement, the influence of reflection on metal workpieces is eliminated. Then, hybrid filtering is used to denoise the image, and finally, the image is recognized. By using the improved Hu moment to perform rough matching on the outer contour, extracting flange holes, and performing local SIFT feature point matching on the holes, violent matching and random consistency are used to reduce mismatched points, thereby identifying the target flange. The experimental results show that the improved Hu moment matching can significantly improve the matching accuracy compared to the original Hu moment matching. |
| keywords:two-hole flange recognition SIFT feature point matching hybrid filtering Hu moment matching |
| 查看全文 查看/发表评论 下载pdf阅读器 |
次