为探究光斑图像在飞秒激光加工过程中的实际意义, 采集等离子体光斑图像并对图像的深度进行提取及分析。本文提出了一种基于图像序列的焦点深度提取(Depth Extraction from Focus Sequence)的方法, 可以有效地解决图像的特征点提取与匹配的问题, 得到更加精确的深度信息。首先, 采用主成分分析(PCA)方法和高通滤波结合的方法对输入的序列图像进行预处理。其次, 各向异性热扩散算法来增强空间频率特征, 增加光斑图像与背景区域的灰度反差值, 得到完整的光斑图像, 并对获得的图像进行像素提取, 选择其中级别最好的像素。然后用矩形中值滤波器过滤光斑的像素图像, 得到更加平滑, 噪声更少的图像。最后, 提取图像的深度信息, 并与传统方法得到的结果进行对比分析, 证实此方法优于传统的图片深度的提取方法, 提高了光斑三维形貌精度, 光斑能级层数得到了细化, 能级层数比传统方法增加了125%。

In order to explore the practical significance of spot image in femtosecond laser processing, plasma spot image was collected and its depth was extracted and analyzed.In this paper, a Depth Extraction from Focus Sequence method based on image Sequence is proposed, which can effectively solve the problem of Extraction and matching of image feature points and obtain more accurate Depth information. Firstly, principal component analysis (PCA) and high pass filtering are used to preprocess the input sequence images. Secondly, anisotropic thermal diffusion algorithm is used to enhance the spatial frequency characteristics and increase the gray contrast value between the light spot image and the background area to obtain a complete light spot image. The obtained image is extracted by pixel and the pixel with the best level is selected. Then filter the pixel image of the light spot with a rectangular median filter to get a smoother image with less noise.Finally, the depth information of the image is extracted and compared with the results obtained by the traditional method. It is proved that this method is superior to the traditional method in image depth extraction, improves the accuracy of three-dimensional morphology of the light spot, and refines the number of light spot level layers, which increases by 125% compared with the traditional method.