Journal of The Optical Society of America A-optics Image Science and Vision最新文献

High-speed laser cladding is a critical surface modification technique, where the melt pool state critically determines the quality of the cladding layer. To address the challenges of inaccurate monitoring and the lack of real-time feedback control during the cladding process, this study proposes an improved YOLOv5-based object detection model, named YOLO-MPID. The model incorporates the Squeeze-and-Excitation (SE) module and the Convolutional Block Attention Module (CBAM) to enhance the extraction of complex edge structures and semantic features in melt pool images. Ablation studies were conducted to evaluate the individual and combined effects of SE and CBAM on detection performance. Comparative experiments under varying laser power conditions demonstrated that YOLO-MPID achieves superior robustness and accuracy compared to the baseline YOLOv5 and other mainstream detection algorithms. Experimental results show that the proposed model achieves a mean Average Precision (mAP at 0.5) of 97.42% and a real-time inference speed of 121.71 FPS. Visual analysis further supports the quantitative findings. In summary, YOLO-MPID provides an effective solution for real-time melt pool state detection and quality control in practical industrial scenarios, offering robust technical support for advancements in image science and optical process control.

Conventional structured light recognition methods rely on spatially resolved imaging. These systems often suffer from low frame rates, sensitivity to alignment, and high computational demands. Such limitations hinder their use in real-time and scalable applications. Here, we demonstrate a novel approach, to our knowledge, for structured light recognition by mapping two-dimensional spatial features onto one-dimensional temporal speckle sequences. This is achieved using a single-pixel detector that captures temporal fluctuations in speckle patterns produced by a rotating diffuser. We demonstrate that optimal mapping occurs when the detector size is equal to or greater than the average speckle grain size, ensuring effective mapping of spatiotemporal speckle dynamics. Utilizing this principle, we successfully recognize Laguerre-Gaussian, Hermite-Gaussian, and intensity-degenerate perfect vortex beams via a support vector machine classifier. The recognition model exhibits >99% accuracy and is robust to atmospheric turbulence, strict optical alignments, or symmetry-breaking optics. Furthermore, we demonstrate a proof-of-concept of the proposed method by establishing a free-space optical communication channel. Employing 16 orbital angular momentum superposition states utilizing a 4-bit binary amplitude switching scheme, we achieve a bit error rate of 0.001. This work presents a scalable, low-latency, and computationally efficient method for real-time structured light recognition, offering significant potential for next-generation optical communication and sensing systems.

The intrinsic connection between enpolarization and depolarization in linear polarimetric transformations becomes evident from the fact that, in general, the measured Mueller matrix-representing integral effects over temporal, spatial, and spectral domains-exhibits both behaviors in an inseparable manner. This entanglement prevents the unambiguous assignment of enpolarizing and depolarizing effects to distinct serial components of the Mueller matrix. In particular, the diattenuators arising in serial decompositions typically display polarizance or diattenuation properties that differ from those of the original system, and a similar situation occurs for the depolarizing component. As for the characterization of retardance, it requires the introduction of entrance and exit retarders, whose definition must rely on appropriate and physically meaningful conventions. This work discusses the problem of decomposing a general Mueller matrix into an equivalent serial system in which the enpolarizing-depolarizing and retarding properties are isolated in separate components. Based on the algebraic structure of Mueller matrices, the proposed solution enables the identification of a set of 16 parameters that independently characterize the system's enpolarizing, depolarizing, and retarding features. The interpretation and physical significance of these parameters are analyzed and discussed.

In this paper, several 2D non-paraxial light fields near the initial plane (sinc beam, Gaussian beam, and Hankel beam) are considered. These TE-polarized beams are shown theoretically and numerically to have a reverse canonical energy flux in the initial plane. The negative value of the longitudinal projection of the canonical flux vector means that the longitudinal derivative of the phase is negative, and hence, the local wave vector is directed against the positive direction of the optical axis. The reverse flux occurs in the presence of evanescent waves. Although running along the initial plane, evanescent waves themselves do not transfer energy along the optical axis perpendicular to the initial plane; they participate in the formation of the reverse flux. Near the initial plane, the amplitude of the light field consists of a linear combination of amplitudes of the propagating and evanescent waves. Therefore, evanescent waves participate in the formation of the phase function of the complex amplitude of the entire field and affect the sign of the longitudinal phase derivative. For a sinc beam, the backflow occurs in the first intensity side lobes, where the standing evanescent wave has odd antinodes.

After previously highlighting time dilation and time reversal according to the refractive index value of the crossed medium, the temporal behavior at the observer place is studied when this refractive index varies as a function of time. Assuming a sinusoidal variation, two modes of progression with a stage of switching between these two modes are differentiated. The mode alternating time dilation and time reversal corresponds to the temporal variation with an alternating progression. The mode slowing down the proper time in the first step and accelerating it without reversing it in the second step corresponds to the temporal variation with a positive progression. The vision of an observer looking through a medium with a variable refractive index reveals directional variations of time and the simultaneity of several events in the scene. The determination of the number of simultaneous events as a function of observed time also indicates if the events are oriented to the future or to the past. Each orientation shows an acceleration zone, a deceleration zone, and a constant zone. These zones are determined by the value of the derivative of the observed time. The frozen time placed on the line of intersection between the future and the past corresponds to the derivative equal to zero. At that moment, the vision of an infinite number of events appears during a dot in observed time. This vision reveals a temporal singularity. The observed simultaneity can no longer be identified by an event number but by the duration of the proper time within the limits of frozen time. All the results obtained in this study were verified and validated, confirming the directional variations of the observed time and the simultaneity of events with the temporal singularity.

Intensity fluctuations quantified by the scintillation index are evaluated in jet engine exhaust turbulence when higher-order laser modes are used in optical wireless communication links. The jet engine exhaust turbulence power spectrum, modified by low-pass and high-pass filters, is employed. Intensity fluctuations are evaluated against the link length, structure constant, wave number (inverse of wavelength) (i.e., against turbulence strength), source size, and jet engine exhaust turbulence parameters. It is found that higher-order laser modes are better at mitigating the scintillations. Jet engine exhaust turbulence parameters are found to affect scintillations substantially.

We develop a prescription for constructing split-step models for computationally efficient numerical propagation of optical fields through distributed atmospheric phase screens in imaging geometries with both an aperture stop and a field stop. We use phase-space optics to show that our prescription yields efficient spatial sampling values, efficient in the sense that they closely match the minimum space-bandwidth product requirements for sampling a generic speckled field that passes through both stops. We also compute the computational complexity for propagating fields through our split-step models using fast Fourier transform methods. Compared with a typical split-step model, our efficient prescription provides more than a 6× reduction in computational complexity with excellent fidelity.

We introduce a radially polarized partially coherent beam characterized by a tailored non-uniform correlation function, termed the radially polarized hyperbolic sine non-uniform coherent (RPHSNUC) beam. We establish the validity conditions for generating this physical optical field. Employing the ordinary Huygens-Fresnel principle, we derive analytical expressions for the spectral intensity and spectral degree of polarization in free space and explore the beam's propagation properties through numerical simulations. Our results demonstrate that RPHSNUC beams preserve their dark hollow core and radial polarization during propagation in free space, while exhibiting a distinctive self-focusing behavior. These findings suggest potential applications in free-space optical communications, polarization-sensitive imaging, and optical trapping.

Concentric optical cameras are widely used in the field of space target detection; the fiber-optic panel is positioned between the concentric objective lens and the sensor, but the presence of defects in the fiber-optic panel will exacerbate the non-uniformity of the image. In this paper, we obtain flat-field and dark-field images and establish the fiber-optic panel mask by designing the calibration experiments. The pixel-scale multipoint correction method is used to correct the non-uniformity of the image after bilinear interpolation corrects the values of the mask points. With the ability to reduce image non-uniformity from 6.3494% to 3.5179% for low irradiance regions and from 1.8826% to 0.3009% for high irradiance regions, this method can improve image quality and analytical reliability under a variety of lighting conditions. The method provides a reference for the image quality optimization of optical imaging systems based on fiber-optic imaging devices, which improves the imaging accuracy and data reliability of concentric optical cameras.

We consider the Goos-Hänchen and Imbert-Fedorov shifts arising from the reflection and transmission of partially coherent, partially polarized beams. For the specific case of an electromagnetic multi-Gaussian Schell-model source, compact expressions are obtained for the spatial and angular versions of the beam shifts, and the influence of the coherence and polarization of such a beam on these shifts is then investigated. It is found that the angular part of these shifts can be significantly enhanced or suppressed by the correlation structure and coherence width of the incident beam. In addition, the state and degree of polarization are not crucial for GH shifts. Both GH effects can be observed even in the completely unpolarized case. However, the IF shifts are more affected by the presence of the unpolarized part of the source.