Abstract:In the closed-loop detection system of the LAMOST, the lens of the fiber view camera system must focus the end of the optical fiber to accurately acquire the fiber position. It is difficult to evaluate the fibers image with a very small proportion of the image with the traditional autofocus algorithm, whether it is recognition of the front illuminated by the fiber ceramic ferrules, or the light-emitting spot of the fibers in the black background, that is, the recognition of the back illumination of the fibers. In this paper, we propose an autofocus determination method for the LAMOST closed-loop control under front and back illumination conditions. Under the condition of front illumination, to greatly reduce the calculation time, the system first pre-recognizes the focusing target through the Faster R-CNN and then uses an optimized contrast algorithm to evaluate the image definition of the corrected focusing target ROI. Then, the algorithm is compared with Tenengrad gradient and Laplacian gradient calculations using the Sobel operator and Laplacian operator in OpenCV. The results show that this method takes only one-ninth of the time required by the other methods to obtain the same accuracy. Under the condition of back illumination, we use the average number and average brightness of spot pixels as the evaluation basis of image sharpness. This method can complete the evaluation of image sharpness in the process of spot recognition and provide initial data for the subsequent detection for closed-loop control. The focus accuracy of the camera is of great significance for thousand-fiber metrology, which will have an important influence on the accuracy of astronomical observations. These focusing methods not only play an important role in the closed-loop control of the LAMOST but also apply to the focusing of closed-loop detection systems of other multi-target optical fiber spectral astronomical telescopes.

Learning to High-Performance Autofocus Microscopy with Laser Illumination

Single-shot Autofocusing in Light Sheet Fluorescence Microscopy with Multiplexed Structured Illumination and Deep Learning

High-accuracy Differential Autofocus System with an Electrically Tunable Lens.

Machine learning guided rapid focusing with sensor-less aberration corrections.

Design and Optimization for Microscope Autofocus Systems Based on the Analysis of Video Signals

Study on Auto—focusing of Laser Scanning Confocal Microscopy

A High Precision Laser-Based Autofocus Method Using Biased Image Plane for Microscopy

Autofocus methods based on laser illumination

Fast Autofocus of Microscopy Images Based on Depth-from-Defocus

Learning to autofocus in whole slide imaging via physics-guided deep cascade networks

Low-cost whole slide imaging system with single-shot autofocusing based on color-multiplexed illumination and deep learning

Image-Based Auto-Focus Microscope System with Visual Servo Control for Micro-Stereolithography

High-robustness autofocusing method in the microscope with laser-based arrayed spots

Lensfree Auto-Focusing Imaging with Coarse-to-fine Tuning Method

Deep Learning-Based Single-Shot Autofocus Method for Digital Microscopy

Autofocus Methods For Automated Microscopy

Single-shot autofocusing of microscopy images using deep learning

Efficient high-resolution fluorescence projection imaging over an extended depth of field through optical hardware and deep learning optimizations

Robust autofocusing method for multi-wavelength lensless imaging.

Research on Autofocus Recognition of the LAMOST Fiber View Camera System under Front and Back Illumination

A Single-Shot Autofocus Approach for Surface Plasmon Resonance Microscopy