This PDF file contains the front matter associated with SPIE Proceedings Volume 12225, including the Title Page, Copyright information, Table of Contents, and Conference Committee listings.

The National Ignition Facility (NIF) employs 192 laser beams to achieve inertial confinement fusion by irradiating a mm scale fusion target. The optical Thomson scattering (OTS) laser is being deployed to probe the target and understand the target implosion physics. Centroid based approach is one of the common approaches for detecting the position of normal Gaussian beams within the OTS laser for beam alignment. Recently, we reported some results of aligning such a beam in 2021, where a pattern matching technique such as matched filtering was used. However, when we defined the template, it included a very high background noise. The correlated noise resulted in an artificial stability when the template was applied on a set of images taken at the same position in quick succession. However, when applied for alignment on different days, the presence of noise had a lesser effect as it made the noise more uncorrelated. In this paper, we re-evaluate this same dataset as published in 2021. We show that the performance of a matched filtering followed by a weighted centroid can overcome distortions appearing in the beam image and is capable of tracking the pattern motion reliably. This paper aims to explain some of the conclusions reached in the previous work while presenting a better approach.

Photonic measurement of microwave signals promises broader bandwidth, higher resolution, and increased resilience to electromagnetic interference relative to conventional electronic instruments for the same size, weight and power (SWaP). Here we demonstrate a microwave photonic spectrometer based on laser speckle pattern imaging. We modulate the RF signal on a frequency-stabilized CW laser using an electro-optic intensity modulator. The modulated CW laser travels through a 100-meter long high-NA multimode optical fiber before collection and recording on a low noise, high dynamic range silicon camera. An RF tone generator is stepped over the desired operational range of the spectrometer with frequency step size on the order of the desired frequency resolution to record the associated speckle pattern. An unknown signal under test is then generated and recovered from the recorded speckle patterns using regression analysis techniques. The spectrometer has been demonstrated over a 17-GHz frequency range with single tone resolution of better than 5-MHz using a penalized ℓ¹ norm (lasso) regression.

Good health and functional ability are important for individuals to lead fulfilling mental, psychological, and social lives. The diseases such as Dementia causes irreversible damage, decline in cognition, function, and behavior which translates into difficulty in independently performing daily tasks. Studies showed that assessment of Instrumental activities of daily living(IADLs) correlate with humans' cognitive and functional status. Analysis of biomechanical markers such as hand movement/use was done with artificial intelligence(AI). We present an optimized AI algorithm for hand detection in the analysis of egocentric video recordings. This improved AI algorithm is based on a probabilistic approach where hand regions are detected in egocentric videos. They then feed the human functional pattern recognition process. To evaluate the performance of our proposal we use a dataset containing the four functional patterns organized into four classes, based on the prehensile patterns of the hands: strength-precision, and on the kinematics of the instruments: displacementhandling. This work was inspired by a previous work done by our group, where biomechanical markers were analyzed throughout the performance of IADL activities to recognize the human functional pattern. The result of our proposal yielded an accuracy of 87.5% in recognizing strength-precision and displacement-handling movement patterns when evaluating the test database with information from Segmented and Not-Segmented videos. This resulted in a single video that changed its classification ratio between the two subsets. This can be of great potential in the development of technological tools for the creation of an automated model to support the diagnosis of early Alzheimer's disease.

This paper presents the implementation of a driving assistance algorithm based on semantic segmentation. The proposed implementation uses a convolutional neural network architecture known as U-Net to perform the image segmentation of traffic scenes taken by the self-driving car during the navigation, the segmented image gives to every pixel a specific class. The driving assistance algorithm uses the data retrieved from the semantic segmentation to perform an evaluation of the environment and provide the results to the self-driving car to help it make a decision. The evaluation of the algorithm is based on the frequency of the pixels of each class, and on an equation that calculates the importance weight of a pixel with its own specific position and its respective class. Experimental results are presented to evaluate the feasibility of the proposed implementation.

Experimental platforms are necessary to evaluate the performance of algorithms in different navigation scenarios. Physical platforms require materials and time to create a single experimental scenario. This approach becomes impractical for exhaustive evaluation in different scenarios because of the prohibitive increase of resources, time, and space. This paper proposes a multi-projector system to mitigate the time and cost by projecting dynamically designed scenes for vehicle navigation experiments. Theoretical principles of perspective projection and mosaicing are reviewed. The dynamic platform is presented for different vehicle navigation cases. The results show that the proposed approach is feasible for vehicle navigation evaluation.

Vision systems have become a promising feedback sensor for robot navigation due to their ability to extract meaningful scene information. In this work, a multicamera system is proposed to estimate the position and orientation of an omnidirectional robot. For this, three calibrated devices (two smartphones and a webcam) are employed. Also, two badges of different colors are placed on the omnidirectional robot to detect its position and orientation. The obtained pose information is used as feedback for the robot trajectory controller. The results show that the proposed system is a useful alternative for the visual localization of ground mobile robots.

Three-dimensional object reconstruction is an essential task in many computer applications. In essence, it consists of firstly estimating the disparity of all corresponding points of an observed scene from a pair of stereo images and then determining the depth map of the scene by triangulation from the estimated disparity. Conventionally, the baseline is fixed in general-purpose stereo cameras. This can limit the resolution and robustness of the three-dimensional reconstruction. In this work, a multi-baseline stereo vision approach for three-dimensional object reconstruction is presented. The mathematical principles of multi-baseline stereo vision are provided. Additionally, experimental results of three-dimensional object reconstruction are presented and discussed in terms of objective measures.

Spatial transfer functions of a volume reflection grating written in photorefractive lithium niobate have been derived and simulated. The Helmholtz equation has been solved analytically under the assumptions that the formed grating is lossless, only two waves exist in the solution, and the wave envelopes are slowly varying functions of space. The grating is assumed to be un-slanted, and the writing and reading wavelengths are equal to 514.5 nm in vacuum. Upon being probed in the reflection geometry, the reflected field experiences low pass spatial filtering, and the transmitted field experiences high-pass spatial filtering. The respective strength of this filtering is determined by the index modulation depth of the reflection grating. Simulation results show this 2D image processing capability of volume reflection gratings in photorefractive materials. In addition, a system has been constructed to test these results experimentally, with initial experimental evidence of spatial filtering.

Modern advances in optical metrology and computer vision have provided an unprecedented ability to generate a wide variety of 3D digital content. The mouse, trackpad, and touch screens are typical 2D interactive interfaces of digital content. However, such interfaces are restrictive to manipulate 3D content such as models, object scans, and environments. In this work, a 3D pointer based on stereo vision to interact virtually with digital 3D objects is proposed. The theoretical principles and the experimental calibration procedure are provided. The proposed 3D pointer is evaluated experimentally by simple interaction routines with objects reconstructed by fringe projection profilometry.

Analysis of two-dimensional spatial data is important for aerial surveys of crops and soils, reconnaissance, medical diagnosis, geographical information systems, and many other domains. Partitioning of images is helpful in the processing and analysis of spatial data. Investigations have shown that splitting an image into sub-images and compressing each sub-image using fewer calculations leads to a faster and more efficient method for the compression of the main image. The relationship between partitioning, spatial information, and the ease of compressing is explored in this paper.

This paper presents a comparison between the implementation of different convolutional neural network models varying the usage of pooling layers to address the problem of hiragana character classification. This study is focused on understanding how the selection and usage of different pooling layers affect the accuracy convergence in a model. To assess this situation eight models were tested with different configurations and using minimum pooling, average pooling, and max pooling schemes. Experimental results to validate the analysis and implementation are provided.

Camera pose estimation is an essential task in many computer vision applications. A widely used approach for this task is given by the specification of several corresponding points in a pair of captured input and reference images. The effectiveness of these methods depends on the accuracy of the specified points and is very sensitive to outliers. This work presents an iterative method for camera pose estimation based on local image correlation. The pose of the camera is estimated by finding a homography matrix that produces the best match between local fragments of the reference image constructed around the specified points and their corresponding projective transformed fragments of the input image when using the estimated homography. The performance of the proposed method is tested by processing synthetic and experimental images.

Numerical simulation to calculate the free spectral range scans (FSR scans) of laser resonators is a computationally intensive task. OSCAR is a well-established Matlab toolbox that enables for such simulations based on Fourier optics. Any arbitrary discrete complex electromagnetic input fields as well as misalignment or mismatching of resonators can be considered in the FSR simulation. Unfortunately, it currently only features CPU based calculations on one or more CPU cores. However, the computational cost increases exponentially with increasing lateral resolution of the complex electromagnetic fields. In addition, only a limited number of roundtrips can be carried out in an acceptable computation time, which limits the applicability only to low finesse resonators. Due to good parallelizability of the FSR scan calculation, this numerical computation is very well suited for modern graphics cards, which are outstanding in performing many calculations in parallel. This paper introduces the extension of FSR scan simulations on modern graphics cards (GPUs) within the OSCAR Toolbox. First, a statistical analysis is provided, that presents the massive performance improvement compared to CPU computations. Subsequently, the disadvantages in the form of memory limitations of GPUs are outlined. Therefore, generally valid data is presented, from which a trade-off between lateral resolution of the complex electromagnetic fields and the number of roundtrips to be performed can be derived. In conclusion, the great potentials of new applications are highlighted, which were previously not feasible. Any code of this GPU implementation discussed in this paper has been integrated into the OSCAR Matlab Toolbox and is made available open source on GitHub.

This paper presents the implementation of a convolutional neural network employing two different malware datasets. These datasets are converted to images, processed, and resized to 64x64. Through image processing, the convolutional neural network can accurately classify the types of malware families in the datasets. Experimental results to validate the analysis and implementation are provided; they were specifically made to show the proposal’s effectiveness and efficiency.

In this paper we’ll describe new techniques of information hiding in visual containers. Presented methods will be connected with the carrier features, so secret distribution will be dependent from the content and pixel values. In traditional secret hiding solutions the way of information distribution is not dependent on the content or container’s features. Such approaches simply modify spatial or frequency parameters, without consideration of local carrier parameters. Such parameters can also determine the way of secret hiding and allow to split information over container in different manner, especially when we consider containers with different graphical features.