基于图像拼接的微小零件视觉检测方法

针对毫米级微小尺寸零件的高准确度检测问题,提出了一种基于图像拼接技术的视觉检测方法,即利用亚像素微小位移结合相位相关的自动拼接方法。以两幅图像的拼接为例,这种方法首先对第一幅图像利用三次样条插值,进行亚像素微小位移;然后用处理后的第一幅图像与第二幅图像进行相位相关处理,计算两者的相关功率谱函数分布图;再根据分布图中功率谱最大值求出两幅图像的亚像素配准关系;最后根据配准关系把两幅图像拼接起来。实验表明,使用这种拼接方法,准确度可达到亚象素级,并具有抗噪声、易实现自动化等特点。

Vision Measurement Method Based on Image Mosaicing for Micro-part

In order to increase the detecting accuracy of micro-parts,a vision detecting method lying on the image mosaicing has been brought forward in this paper,which is based on subpixel displacement interpolation and phase correlation.Two images have been taken as a sample.Cubic interpolation has been used to displace one image at first;cross power spectrum of the two images has been computed later;subpixel registration has been found out then,based on the maximum data of cross power spectrum distributing function;image interpolation is used to end the job.The experiment shows that: high accuracy,anti-noise and automatization can be achieved by this method.