为实现对飞秒激光烧蚀的光斑阵列进行准确测量,构建微视觉系统进行图像的采集、处理。由于微视觉系统视场有限,单帧图像无法反映光斑阵列全貌,需对多帧图像进行拼接处理。首先,利用微视觉远心镜头对飞秒激光烧蚀的光斑列阵图像进行采集,获得多幅光斑阵列图像; 其次,运用Harris及RANSAC算子并加入ANMS非极大值抑制,对多幅光斑阵列图像进行配准与拼接,并经实验验证了采用此图像拼接方法能获得较好的光斑阵列整体图像。最后,运用SSIM评价法对拼接后图像与原图像进行对比,证明了拼接效果的准确性,能达到对原图像89.795 435%的还原,对测量大视场微结构图像有重要意义。

In order to accurately measure the spot array of the femtosecond laser ablation, the image acquisition and process of micro visual system is constructed. Because of the limited view of the micro visual system, the single frame image cannot reflect the whole image of the ablation spot array and splices the multi-frame images. Firstly, The spot array images of femtosecond laser ablation are collected by using micro visual telecentric lens, and multiple spot array images are obtained. Secondly, using Harris and RANSAC operators and adding ANMS non-maximum suppression, multiple spot array images are registered and stitched. It is verified by experiments that the image mosaic method can obtain a better global image of spot array. Finally, the SSIM evaluation method is used to compare the stitched image with the original image, which proves the accuracy of the stitching effect and can achieve the restoration of 89.795 435% of the original image, which is of great significance for measuring large field of view microscale structure images.