基于深度学习的模糊激光三维图像重建研究  

Research on Fuzzy laser 3D image reconstruction based on depth learning

作　　者：徐慧[1] 余晓丽[1] XU Hui;YU Xiaoli(Nanchang Institute of Technology,Nanchang 330044,China)

机构地区：[1]南昌理工学院,南昌330044

出　　处：《激光杂志》2022年第12期108-112,共5页Laser Journal

基　　金：江西省教育厅科技项目(No.GJJ191012)。

摘　　要：重建作为改善激光三维图像的重要途径,针对传统方法的激光三维图像重建误差大,重建耗时长等局限性,以改善激光三维图像重建效果为目标,提出基于深度学习的模糊激光三维图像重建方法。首先分析了激光三维图像的研究进展,找到各种激光三维图像重建方法的不足,然后采集激光三维图像,采用小波变换算法对激光图像进行去模糊处理,提升激光三维图像质量,并引入深度学习方法设计模糊激光三维图像重建模型,并与其他模糊激光三维图像重建方法进行了性能对比。结果表明,相对于其他激光三维图像重建方法,深度学习方法明显提升了模糊激光三维图像重建效果,激光三维图像信噪比提升很高,重建速度更快,时间控制在20 ms以内,重建精度超过92%,获得了更优的激光三维图像重建效果。Reconstruction is an important way to improve laser 3D image. In view of the limitations of traditional methods, such as large error and long reconstruction time, in order to improve the effect of laser 3D image reconstruction, a fuzzy laser 3D image reconstruction method based on depth learning is proposed. Firstly, the research progress of laser three-dimensional image is analyzed, and the shortcomings of various laser three-dimensional image reconstruction methods are found. Then, the laser three-dimensional image is collected, the wavelet transform algorithm is used to de blur the laser image, so as to improve the quality of laser three-dimensional image, and the depth learning method is introduced to design the fuzzy laser three-dimensional image reconstruction model, the performance is compared with other fuzzy laser 3D image reconstruction methods. The results show that compared with other laser 3D image reconstruction methods, the depth learning method significantly improves the effect of fuzzy laser 3D image reconstruction, the signal-to-noise ratio of laser 3D image is increased by about 20%, the speed of laser 3D image reconstruction is faster, the time is controlled within 20 ms, and the accuracy of laser 3D image reconstruction is more than 92%, The laser three-dimensional image reconstruction effect is obtained.

关 键 词：激光技术 三维图像 图像信噪比 重建精度 深度学习方法 

分 类 号：TN304[电子电信—物理电子学]