KEYWORDS: Super resolution, Education and training, RGB color model, Data modeling, Transformers, Performance modeling, Image enhancement, Visual process modeling, Aliasing, Gallium nitride
Many deep learning-based image super-resolution models exist to effectively up-sample images, with the most notable and reliable architectures being Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and Generative Adversarial Networks (GANs). To date, model benchmarking has been made only with the same architecture type or only with certain datasets that could potentially be beneficial to the proposed models. In this paper, we present the first-known comparison of state-of-the-art super-resolution models, namely, SwinIR, EDSR, Swin2SR and Real- ESRGAN, to serve as a reference baseline for future applications where the modelling complexity, frame rates and overall super-resolution accuracy is of concern. The experiments were conducted by reproducing the models entirely by following the training procedures highlighted in their original paper. Then, we performed the evaluations on the conventional image super-resolution test sets, namely, Set5, Set14, BSD100, Urban100, T91 and Manga109. Our experimental results show that each model has their respective tradeoff between the number of measures taken to suppress the super-resolution artifacts and achieve a higher super-resolution accuracy and the overall model processing times, such as the model convergence speed and their respective frame rates.
Single-pixel camera is developed to mitigate the constraints faced by the conventional cameras especially in invisible wavelengths and low light conditions. Nyquist–Shannon theorem requires as many measurements as the image pixels to reconstruct images flawlessly. In practice, obtaining more measurements increases the cost and acquisition time, which are the major drawbacks of single-pixel imaging (SPI). Therefore, compressive sensing was proposed to enable image reconstruction with fewer measurements. We present a design of sensing patterns to obtain image information by utilizing spatially variant resolution (SVR) technique in SPI. The proposed method reduces the measurements by prioritizing the resolution in the region of interest (ROI). It successfully achieves the programmable imaging concept where multiresolution adaptively optimizes the balance between the image quality and the measurements number. Results show that SVR images can be reconstructed from significantly fewer measurements yet able to achieve better image quality than uniform resolution images. In addition, the SVR images can be further enhanced by integrating the dynamic supersampling technique. Consequently, the concerns of image quality, long acquisition, and processing time can be addressed. The proposed method potentially benefits imaging applications where the target ROI is prioritized over the background and most importantly it requires fewer measurements.
Recently, non-line-of-sight (NLOS) imaging is a novel optical computational imaging technology developed. In NLOS imaging, objects are reconstructed by analyzing information carried in the reflected light. Considering the flaws in NLOS such as information loss and weak target detection due to diffuse reflection of light, developing a stable and effective imaging system is nontrivial. Therefore, in this paper, a 2D object inspection system based on compressed sensing and Orthogonal Matching Pursuit (CSOMP) approaches for effective NLOS imaging is proposed. This will not only alleviate the difficulty in collecting object information in a low light environment but also make the reconstruction process with a few measurements. Specifically, the single-pixel camera based on Digital Micromirror Device (DMD) is introduced to fully exploit the programmable pattern information and the correlation between the measured intensity data for image renovation. To reduce acquisition time, a high-efficiency single-pixel detector is employed that obviates the need for mechanical scanning. Also, in the image reconstruction stage, the residuals of the measurement matrix are obtained through OMP. The proposed method combines with the CS theory; thus, the target can be rebuilt with fewer measurements. Experimental results on NLOS images illustrate that the proposed imaging scheme reconstructed the image exactly similar to the real object.
Performance of a range gated system is strongly affected by the laser, sensor, target, and atmospheric parameters. This paper performs a theoretical analysis to investigate the influence of multiple factors on range gated reconstruction. The effects of several factors are discussed based on the operating principle of range gated reconstruction, fundamental of radiant energy, signal to noise ratio (SNR), and bidirectional reflection distribution function (BRDF) models. The presented findings establish a comprehensive understanding of the influence factors in range gated reconstruction which are of interest to various applications and future improvement works to perform accurate range correction and compensation.
Recently, multi-layer surface profiling and inspection has been considered an emerging topic that can be used to solve various manufacturing inspection problems, such as graded index lenses, TSV (Thru-Silicon Via), and optical coating. In our study, we proposed a gated wavefront sensing approach to estimate the multi-layer surface profile. In this paper, we set up an experimental platform to validate our theoretical models and methods. Our test bed consists of pulse laser, collimator, prism, well-defined focusing lens, testing specimen, and gated wavefront sensing assembly (e.g., lenslet and gated camera). Typical wavefront measurement steps are carried out for the gated system, except the reflectance is timed against its time of flight as well as its intensity profile. By synchronizing the laser pulses to the camera gate time, it is possible to discriminate a multi-layer wavefront from its neighbouring discrete layer reflections.
General precursors and growth model of Laser Induced Damage (LID) have been the focus of research in fused silica material, such as polishing residues, fractures, and contaminations. Assuming the absorption due to trapped material and mechanical strength is the same across the surfaces, various studies have shown that the LID could be minimized by reducing the light field intensification of the layers upon the laser strikes. By revisiting the definition of non-ionising radiation damage, this paper presents the modelling work and simulation of light intensification of laser induced damage condition. Our contribution is to predict the LID growth that take into various factors, specifically on the light intensification problem. The light intensification problem is a function of the inter-layer or intra-layer micro-optical properties, such as transmittance and absorption coefficient of the material at micro- or sub-micro-meter range. The proposed model will first estimate the light propagation that convoluted with the multiply scattering light and subsequently the field intensification within the nodule dimension. This will allow us to evaluate the geometrical factor of the nodule effect over the intensification. The result show that the light intensification is higher whenever the backscattering and multiple scattering components are higher due to its interference with the incoming wave within its coherency.
Range gated imaging is a remote sensing acquisition which involves the emission of a laser pulse and an intensified
camera to gate the reflected laser pulse. Range accuracy has always been an issue especially when a highly accurate
reconstructed model is expected as the final outcome. The reflected pulse profile and pulse instability are among the
issues that affect the range accuracy when a general solution such as constant offset is not applicable. In this paper, a
study to estimate a more accurate model for the reflected pulse profile has been investigated through experiments. T
Location-Scale model has been proposed to replace the Gaussian model as the general assumption for range-gated image
formation model. The improvement on range accuracy which is around 0.3% has been verified through simulation based
on the acquired samples. The series of range-gated images can be reconstructed into a three-dimensional depth map
through range calculation. This can be used in the subsequent range reconstruction works.
Recently, semiconductor manufacturers have been striving for high speed, large scale multi-layer wafer surface measurement. In this paper, we propose a novel technique in multi-layer wave-front sensing. The measurement uses a gated camera in pico second shutter that can be synchronized to a pico second laser pulse, up to μm accuracy. Subsequently, we propose a compensation technique using time-of-flight wave-front sensing to reconstruct the multilayer surfaces using our proposed gated imaging technique.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.