您的位置：首页 > 论文页面
基于图像深度估计的微位移检测系统
发表时间：2016-11-30 浏览量：1781 下载量：361
| 全部作者： | 袁韬，黄涛，陈文 |
| 作者单位： | 武汉理工大学信息工程学院 |
| 摘 要： | 提出一种基于曲线拟合（curve fitting）的计算机视觉测距方法，并实现了基于图像深度估计的微位移检测系统，选取适用于显微图像的去噪方法和图像清晰度评价函数，以基于曲线拟合的测距算法换算实际的汽车阀阀芯位移，位移检测的误差率为0.92%，清晰度辨识精度为120×103/μm，最小辨识距离为20 μm. 研究不同的去噪滤波方法对Brenner 清晰度评价函数的影响，得出均值滤波既可以保证较好的图像去噪效果，同时也能够在一定程度上提高清晰度评价函数的清晰度辨识精度。在适用于显微环境下的清晰度评价函数选取中，选取具有代表性的灰度方差值、能量梯度函数、Brenner 函数、Entropy 熵函数、Memmay 函数、Roberts 函数和Vollath 函数这7 种图像清晰度评价函数，对于这7 种清晰度评价函数在满足理想清晰度评价函数特性和达到20 μm 清晰度辨识精度上进行相应的实验验证，结果表明，Brenner 函数在均值滤波后达到了预期要求。 |
| 关 键 词： | 计算机应用；图像深度估计；测距算法；曲线拟合 |
| Title： | Detection system of micro displacement based on image depth estimation |
| Author： | YUAN Tao, HUANG Tao, CHEN Wen |
| Organization： | School of Information Engineering, Wuhan University of Technology |
| Abstract： | In this paper, a computer vision distance measurement method based on curve fitting is proposed, the micro displacement detection system based on image depth estimation was designed. We selected suitable image denoising method and image clarity evaluation function to obtain auto valve spool displacement measurement method based on curve fitting to convert the actual displacement, and error detection rate was 0.92%, the accuracy of identification for the 120×103/μm definition, identification of the minimum distance for 20 μm. After studied the influence of different denoising methods on the Brenner resolution evaluation function, we learned that the mean filter could ensure the effect in image denoising, meanwhile, we could improve the accuracy of clarity evaluate function. In the selection of Definition evaluation function under the microscope environment, we chose seven clarity evaluate functions of image, including Gray variance, Energy gradient, Brenner, Entropy, Memmay, Roberts and Vollath. In order to verify whether the seven kinds of clarity evaluation function can meet the characteristics of the ideal definition evaluation function and achieve the 20 μm definition of identification accuracy, we performed verified by experiments, and we got the result that Brenner function achieve the expected requirements in Mean filter. |
| Key words： | computer applications; estimation of image depth; method of displacement measurement; curve fitting |
| 发表期数： | 2016年11月第22期 |
| 引用格式： | 袁韬，黄涛，陈文. 基于图像深度估计的微位移检测系统[J]. 中国科技论文在线精品论文，2016，9（22）：2296-2303. |
2
评论数 0请您登录
暂无评论
下载全文
多维论文