线阵相机场曲复原及在异纤检测中的应用

为了消除相机的场曲给成像带来的空变模糊,提高图像分辨率和检测精度,针对在工业检测中广泛使用的线阵相机,提出了一种消除场曲影响的一维图像复原方法。在分析了空变模糊矩阵结构的基础上,通过估算部分区域的点扩散函数,再由偏移和插值得到全像场范围内的模糊矩阵。由模糊矩阵利用约束最小二乘法得到与观测信号无关的复原矩阵。检测过程中,将采集到的图像信号与复原矩阵相乘,获得复原后的图像。采用检测棉流内异性纤维的异纤检测系统,用含有较细异纤的实际数据对方法进行验证。结果表明,本方法在提高图像边缘处分辨率的同时增强了异常点与背景的差异,异常比增加了10%以上,为进一步提高检测精度创造了条件。

Restoration of field curved image from line camera and its applications in foreign fiber detecting

To eliminate the spatially variant blurs caused by a field curvature from a camera and to improve the resolutions and the measuring precisions of images, a 1D image restoration method was proposed to overcome the influence of field curvature on an image from the camera used in industrial inspections. On analysis of the spatially variant blur matrix, the Point Spread Functions(PSFs) of a partial region were estimated and then the blur matrix of whole image field was computed by shifting and interpolation. On the basis of the blur matrix, the restored matrix which was unrelated to the observed signal was obtained by the constrained least square algorithm. In inspection, the restored image could be obtained by multiplying the acquired signals with restored matrix. This method has been applied in foreign fiber detecting systems and verified by real images containing fine fibers. Obtained results show that this method can improve the image resolution and enhance the difference between background and irregular points by 10%. Therefore, it improves the detecting accuracies for images.