Tao Jin

The theory of retinal peripheral defocus currently represents the prevailing paradigm for myopia prevention and control. Research studies on peripheral refraction indicate that peripheral defocus is significantly associated with the management and progression of myopia. A series of animal and clinical studies conducted over the last century have substantiated this theory. Accurate and rapid measurement and evaluation of peripheral refractive states using appropriate techniques form the foundation for investigating this theory. Research based on technologies for measuring retinal peripheral refractive errors continues to evolve, with precise detection holding substantial significance in ophthalmology, particularly within optometry. This paper provides a comprehensive overview of the principles underlying the most commonly employed methods for measuring peripheral refraction, addressing their current status in both clinical and commercial contexts, and ranging from subjective to objective measurements, as well as analyzing and comparing various methodologies along with their respective advantages and disadvantages. Special emphasis is placed on widely utilized open automatic optometry systems and Hartmann–Shack wavefront aberration techniques, while also introducing a newly developed measurement method grounded in refractive compensation technology for optical imaging—multispectral refractive topography. Based on these measurement techniques, future directions for research and technological advancements concerning peripheral refractive error are proposed.

Torque can intuitively reflect the working status of mechanical systems, making real-time monitoring and accurate measurement of torque essential for ensuring the safe and stable operation of mechanical devices. By analyzing various existing measurement methods, this paper proposes a torque measurement system based on the photoelectric conversion principle. The system uses an LED light source and a four-quadrant detector to capture the minute deformations of the rotating shaft. To mitigate the adverse effects of temperature and current fluctuations on the LED optical power output, a dual closed-loop feedback control circuit for light intensity and current is designed. This system establishes a correspondence between torque measurements and standard values and uses a particle swarm optimization support vector regression (PSO-SVR) algorithm to achieve measurement calibration. Experimental results show that the system has the root mean square error (RMSE) of 0.67, the mean absolute percentage error (MAPE) of 2.4%, and the mean absolute error (MAE) of 0.49, demonstrating good performance in torque measurement.

Surface defect detection is a critical task in the manufacturing industry to ensure product quality and machining efficiency. Image-based precise defect detection faces significant challenges due to defects lacking fixed shapes and the detection being heavily influenced by lighting conditions. Addressing the efficiency demands of defect detection algorithms, often deployed on embedded devices, and the highly imbalanced pixel ratio between foreground and background images, this paper introduces a multi-scale fusion staged U-shaped convolutional neural network (DEU-Net). The network provides segmentation results for defect anomalies while indicating the probability of defect presence. It enables the model to train with fewer parameters, a crucial requirement for practical applications. The proposed model achieves an MIoU of 66.94 and an F1 score of 74.89 with lower Params (36.675) and Flops (19.714). Comparative analysis with FCN, U-Net, Deeplab v3+, U-Net++, Attention U-Net, and Trans U-Net demonstrates the superiority of the proposed approach in surface defect detection.

To solve the short working distance and small measurement range of an all-fibre interferometer, we proposed a Mach–Zehnder Fabry–Perot hybrid fibre-optic interferometry system based on sinusoidal phase modulation. In this paper, a low-finesse fibre interferometer with a larger linear operating range for displacement measurement is realised using a self-collimating probe and incorporating a Kalman filter-based phase demodulation algorithm. Through experimental comparisons, it is demonstrated that the interferometer proposed in this paper can effectively reduce the phase delay, compensate for the depth of modulation drift, and correct the error due to parasitic interference introduced by the optical path structure through the algorithm. A linear large measurement working range of 20 cm is realised.

Eye movement research is an established methodology and widely used in fundamental research fields such as psychology or neuroscience. Most research is performed in controlled environments with screen-based eye trackers, but demand for a wearable requirement is on the rise. Yet, a state of the art wearable eye tracking system thus far has not been feasible for subjects with refractive errors; therefore, a trial frame is proposed for researching the eye movement of people with or without refractive errors. Two iris cameras and several infrared radiation LEDs were fixed behind the frame for recording eye images. The images were sent to a host computer using 2.4G wireless, which was also fixed on the frame. The entire electronic system in the trial frame was driven using a battery. Eye position in the images was recognized and segmented by a model that was trained through machine learning. Then the pupil and blinks were detected from the segmented eye area using ellipse fitting and the eye aspect ratio. Experiments showed that the correct rate of pupil and blink detections were 97.24% and 96.14%, respectively. (c) 2023 Optica Publishing Group

To obtain the pupil center of the human eye in infrared images taken by a near-eye device for gaze tracking, pupil and blink detection algorithms are proposed. The eye-detection model and the eye feature point model are trained through the dlib library machine learning, the eye area is segmented, rough positioning of the eye area is realized, the redundant image information is removed, the number of image processing calculations are removed for the subsequent pupil positioning, and the processing time is shortened. In pupil detection, the candidate pupil contours are screened based on the gray information, shape characteristics and other pupil image information to obtain the correct pupil contour information and realize precise pupil positioning. The eye feature model is used to obtain the coordinate of the feature point of the eye, the aspect ratio of the eye is obtained by conversion, and blink statistics are performed. Experiments show that the correct rate of the pupil detection method reaches 97.24%, and the correct rate of blink detection reaches 91.59%.

A straightness heterodyne interferometer is proposed for measuring the straightness of a linear motion stage along its moving direction. The interferometer uses a Koster prism as the beam splitter and combiner, a wedge prism as the straightness sensor, and a wedge mirror as the reflector. The interferometric beams have a completely common optic path and space symmetrical measurement structure. The cross-talk errors, from other degrees of freedom, are eliminated due to the spatial symmetry optical path structure. However, the periodic nonlinearities due to the mixed frequency cannot be eliminated by the optical configuration. To overcome this problem, we derived a periodic nonlinearities correction model. The coefficients of the model are previously estimated by splitting the straightness measurement beam into two beams through a polarization beam splitter placed at 45°. Afterwards, the model can be used to compensate the nonlinearities of the straightness measurement. The results show that the nonlinear error is reduced to the resolution limitation of the straightness interferometer.

The measurement of the diopters of spectacle lenses using an artificial neural network is proposed in this paper. A diopter is measured by obtaining the distances between the spots imaged by a charge-coupled device (CCD) during a Hartmann test. Backpropagation (BP) and a radial basis function (RBF) algorithm are applied to train the BP and RBF neural networks, respectively. A set of − 20 D to 20 D spectacle lenses is used to test the proposed method. When comparing the diopter measurement results of the RBF neural network with those of the BP neural network, the former exhibited higher stability and lower errors. In addition, the errors of diopter measurement are not against the system error measured by an auto-focimeter due to the increase of curvature of spectacle lenses. These results indicated that RBF neural network has better performance in diopter detection, and it can overcome the system error caused by the change of lens curvature in auto-focimeter measurement.