Search Results (38)

In conventional infrared imaging systems, non-uniformity correction typically involves continuously reading correction parameters from double data rate (DDR) memory. For high-frame-rate short-wave infrared imaging systems to achieve real-time non-uniformity correction, it is essential to minimize the reading time of correction parameters. Due to the narrow dynamic range of two-point correction and the large parameter storage required by two-point multi-segment correction, it is difficult to simultaneously achieve good correction performance and short parameter reading time under limited hardware resources. To address the above issues, this paper proposes a real-time non-uniformity correction method suitable for high-frame-rate short-wave infrared images. Based on a field-programmable gate array (FPGA), improvements are made to quadratic polynomial correction through the design of quantization methods for different parameters to enhance storage bit-width utilization; dynamic allocation of bit-widths between parameters to improve correction performance; and ping-pong buffering for DDR reading to avoid the impact of DDR read latency on parameter reading time. The storage size of the improved correction parameters is comparable to that of conventional two-point correction. Experiments were conducted on a hardware system based on the XC7A100T-2FGG484I FPGA. The experimental results show that the average non-uniformity of images after the improved quadratic polynomial correction is 0.4818%, significantly better than 0.5930% after two-point correction and slightly better than 0.4891% after two-point eight-segment correction. Blind pixel compensation was completed simultaneously with the correction. Using a 640 × 512 area array InGaAs short-wave infrared detector, the highest real-time processing frame rate reaches 800 frames per second (FPS). Full article

This paper proposes a collaborative optimization strategy of regenerative braking in heavy-duty electric logistics vehicles under complex driving conditions to improve energy recovery efficiency. Based on the actual operational data of 18-ton electric trucks in the southwestern region of China, three driving scenarios for heavy commercial vehicles are determined via the K-Means clustering algorithm. Key features are extracted using Recursive Feature Elimination and employed to train a Learning Vector Quantization neural network for precise real-time condition recognition. The identified driving condition parameters, including vehicle speed, remaining battery power, and braking force, collectively regulate the intensity of regenerative braking. Simulation results under double-WTVC (World Transient Vehicle Cycle) conditions indicate that the proposed strategy can effectively adapt regenerative braking behavior to diverse road conditions. In comparison with conventional control methods, this approach enhances battery energy recovery efficiency by 5.8% while preventing control discontinuities. Full article

We investigated an exact solution in a conformal invariant Randall-Sundrum 5D warped brane world model on a time dependent Kerr-like spacetime. The singular points are determined by a quintic polynomial in the complex plane and fulfills Cauchy’s theorem on holomorphic functions. The solution, which is determined by a first-degree differential equation, shows many similarities with an instanton. In order to describe the quantum mechanical aspects of the black hole solution, we apply the antipodal boundary condition. The solution is invariant under time reversal and also valid in Riemannian space. Moreover, CPT invariance in maintained. The vacuum instanton solution follows from the 5D as well as the effective 4D brane equations, only when we allow the contribution of the projected 5D Weyl tensor on the brane (the KK-‘particles’). The topology of the effective 4D space of the brane is the projective $\mathbb{R}{P}^3$ (elliptic space) by identifying antipodal points on $S^3$. The 5D is completed by applying the Klein bottle embedding and the ${\mathbb{Z}}_2$ symmetry of the RS model. This model fits very well with the description of the Hawking radiation, which remains pure. We have also indicated a possible way to include fermions. Our 5D space admits a double cover of $S^3$ and after fibering to the $S^2$, we obtain the effective black hole horizon. The connection with the icosahedron discrete symmetry group is investigated. It seem that Bekenstein’s conjecture that the area of a black hole is quantized, could be applied to our model. Full article

For the application of a high-performance shortwave infrared (SWIR) detector, a fully integrated analog front-end (AFE) circuit is proposed in this paper, which includes a readout integrated circuit (ROIC) and a 12-bit/100 kHz two-step single-slope analog-to-digital converter (TS-SS ADC). The ROIC adopts a direct injection (DI) structure with a pixel size of only 10 µm × 10 µm. The column processing circuit uses a passive correlated double-sampling (CDS) circuit to reduce noise and improve dynamic range. The comparator of four inputs in the ADC solves the problem of linearity reduction caused by charge redistribution during coarse quantization. In addition, the current steering digital-to-analog converter (DAC) is used to compensate for the non-ideal characteristics of the switch, which effectively optimizes the differential nonlinearity (DNL) and integral nonlinearity (INL). The AFE is implemented using SMIC 180 nm 1P6M technology. The post-simulation results show that at a power supply voltage of 3.3 V, the AFE has a frame rate of 100 Hz and a full well capacity (FWC) of 2.8 Me⁻. The linearity can reach 99.59%, and the equivalent output noise is 243 µV. The dynamic range is 73.8 dB. Meanwhile, the signal-to-noise distortion ratio (SNDR) and effective number of bits (ENOB) are 68.38 dB and 11.06 bits, respectively. Full article

Gas leaks threaten ecological and social safety. Non-contact infrared imaging enables large-scale, real-time measurements; however, in complex environments, weak signals from small leaks can hinder reliable detection. This study proposes a novel automated methane leak detection method based on infrared imaging and a Gas-Faster Region-based convolutional neural network (Gas R-CNN) to classify leakage amounts (≥30 mL/min). An uncooled infrared imaging system was employed to capture gas leak images containing leak volume features. We developed the Gas R-CNN model for gas leakage detection. This model introduces a multiscale feature network to improve leak feature extraction and enhancement, and it incorporates region-of-interest alignment to address the mismatch caused by double quantization. Feature extraction was enhanced by integrating ResNet50 with an efficient channel attention mechanism. Image enhancement techniques were applied to expand the dataset diversity. Leak detection capabilities were validated using the IOD-Video dataset, while the constructed gas dataset enabled the first quantitative leak assessment. The experimental results demonstrated that the model can accurately detect the leakage area and classify leakage amounts, enabling the quantitative analysis of infrared images. The proposed method achieved average precisions of 0.9599, 0.9647, and 0.9833 for leak rates of 30, 100, and 300 mL/min, respectively. Full article

This study introduces YOLOv13-Cone-Lite, an enhanced algorithm based on YOLOv13s, designed to meet the stringent accuracy and real-time performance demands for traffic cone detection in autonomous formula racing cars on enclosed tracks. We improved detection accuracy by refining the network architecture. Specifically, the DS-C3k2_UIB module, an advanced iteration of the Universal Inverted Bottleneck (UIB), was integrated into the backbone to boost small object feature extraction. Additionally, the Non-Maximum Suppression (NMS)-free ConeDetect head was engineered to eliminate post-processing delays. To accommodate resource-limited onboard terminals, we minimized superfluous parameters through structural reparameterization pruning and performed 8-bit integer (INT8) quantization using the TensorRT toolkit, resulting in a lightweight model. Experimental findings show that YOLOv13-Cone-Lite achieves a mAP₅₀ of 92.9% (a 4.5% enhancement over the original YOLOv13s), a frame rate of 68 Hz (double the original model’s speed), and a parameter size of 8.7 MB (a 52.5% reduction). The proposed algorithm effectively addresses challenges like intricate lighting and long-range detection of small objects and offers the automotive industry a framework to develop more efficient onboard perception systems, while informing object detection in other closed autonomous environments like factory campuses. Notably, the model is optimized for enclosed tracks, with open traffic generalization needing further validation. Full article

As AI-enabled embedded systems such as smart TVs and edge devices demand efficient video processing, Versatile Video Coding (VVC/H.266) becomes essential for bandwidth-constrained Multimedia Internet of Things (M-IoT) applications. However, its block-based coding often introduces compression artifacts. While CNN-based methods effectively reduce these artifacts, maintaining robust performance across varying quantization parameters (QPs) remains challenging. Recent QP-adaptive designs like QA-Filter show promise but are still limited. This paper proposes DRIFT, a QP-adaptive in-loop filtering network for VVC. DRIFT combines a lightweight frequency fusion CNN (LFFCNN) for local enhancement and a Swin Transformer-based global skip connection for capturing long-range dependencies. LFFCNN leverages octave convolution and introduces a novel residual block (FFRB) that integrates multiscale extraction, QP adaptivity, frequency fusion, and spatial-channel attention. A QP estimator (QPE) is further introduced to mitigate double enhancement in inter-coded frames. Experimental results demonstrate that DRIFT achieves BD rate reductions of 6.56% (intra) and 4.83% (inter), with an up to 10.90% gain on the BasketballDrill sequence. Additionally, LFFCNN reduces the model size by 32% while slightly improving the coding performance over QA-Filter. Full article

This study proposes a method to detect double compression in H.265/HEVC videos containing B-frames, a scenario underexplored in previous research. The method extracts frame-level encoding features—including frame type, coding unit (CU) size, quantization parameter (QP), and prediction modes—and represents each video as a 28-dimensional feature vector. A bidirectional Long Short-Term Memory (Bi-LSTM) classifier is then trained to model temporal inconsistencies introduced during recompression. To evaluate the method, we created a dataset of 129 HEVC-encoded YUV videos derived from 43 original sequences, covering various bitrate combinations and GOP structures. The proposed method achieved a detection accuracy of 80.06%, outperforming two existing baselines. These results demonstrate the practical applicability of the proposed approach in realistic double compression scenarios. Full article

Industrial Internet of Things (IIoT) deployments in thermal power plants face significant energy efficiency challenges due to harsh operating conditions and device resource constraints. This paper presents gradient memory double-deep Q-network (GM-DDQN), a lightweight reinforcement learning approach for energy optimization on resource-constrained IIoT devices. At its core, GM-DDQN introduces the gradient memory mechanism, a novel memory-efficient alternative to experience replay. This core innovation, combined with a simplified neural network architecture and efficient parameter quantization, collectively reduces memory requirements by 99% and computation time by 85–90% compared to standard methods. Experimental evaluations across three realistic simulated thermal power plant scenarios demonstrate that GM-DDQN improves energy efficiency by 42% compared to fixed policies and 27% compared to threshold-based approaches, extending battery lifetime from 8–9 months to 14–15 months while maintaining 96–97% PSR. The method enables sophisticated reinforcement learning directly on IIoT edge devices without requiring cloud connectivity, reducing maintenance costs and improving monitoring reliability in industrial environments. Full article

The development of hardware-based cognitive computing systems critically hinges upon the integration of memristor devices capable of versatile weight expression across a spectrum of resistance levels while preserving consistent electrical properties. This investigation aims to explore the practical implementation of a digit recognition system utilizing memristor devices with minimized weighting levels. Through the process of weight quantization for digits represented by 25 or 49 input signals, the study endeavors to ascertain the feasibility of digit recognition via neural network computation. The integration of memristor devices into the system architecture is poised to streamline the representation of the resistors required for weight expression, thereby facilitating the realization of neural-network-based cognitive systems. To minimize the information corruption in the system caused by weight quantization, we introduce the concept of “weight range” in this work. The weight range is the range between the maximum and minimum values of the weights in the neural network. We found that this has a direct impact on weight quantization, which reduces the number of digits represented by a weight below a certain level. This was found to help maintain the information integrity of the entire system despite the reduction in weight levels. Moreover, to validate the efficacy of the proposed methodology, quantized weights are systematically applied to an array of double-layer neural networks. This validation process involves the construction of cross-point array circuits with dimensions of 25 × 10 and 10 × 10, followed by a meticulous examination of the resultant changes in the recognition rate of randomly generated numbers through device simulations. Such endeavors contribute to advancing the understanding and practical implementation of hardware-based cognitive computing systems leveraging memristor devices and weight quantization techniques. Full article

This paper investigates state estimation methods for dynamical systems when model evaluations are performed on resource-constrained embedded systems with finite precision compute elements. Minimum mean square estimation algorithms are reformulated to incorporate finite-precision numerical errors in states, inputs, and measurements. Quantized versions of least squares batch estimation, sequential Kalman, and square-root filtering algorithms are proposed for fixed-point implementations. Numerical simulations are used to demonstrate performance improvements over standard filter formulations. Steady-state covariance analysis is employed to capture the performance trade-offs with numerical precision, providing insights into the best possible filter accuracy achievable for a given numerical representation. A low-latency fixed-point acceleration state estimation architecture for optomechanical sensing applications is realized on Field Programmable Gate Array System on Chip (FPGA-SoC) hardware. The hardware implementation results of the estimator are compared with double-precision MATLAB implementation, and the performance metrics are reported. Simulations and the experimental results underscore the significance of modeling quantization errors into state estimation pipelines for fixed-point embedded implementations. Full article

The beat signal obtained from frequency-modulated continuous-wave (FMCW) radar is a waveform that is corrupted by noise and requires filtering out interference components for frequency calibration. Traditional FFT methods are affected by the fence effect and spectral leakage, leading to a reduction in frequency estimation accuracy. Therefore, an improved double-spectrum-line interpolation frequency estimation algorithm is proposed in this paper, utilizing spectral refinement and phase interpolation. Firstly, the post-FFT spectral signal is refined to narrow the frequency search range and enhance frequency resolution, thereby separating the noise signal. Then, a frequency deviation factor is defined based on the relationship between adjacent phase angles. Finally, the signal’s phase angles are interpolated using the frequency deviation factor to estimate the frequency of the beat signal. Experimental results demonstrate that the proposed algorithm reduces the impact of quantization on the frequency distribution and increases the signal’s noise resistance. The proposed algorithm has a higher accuracy and lower standard deviation compared to the recently proposed algorithm. Full article

Extensive research has been conducted in image forensics on the analysis of double-compressed images, particularly in the widely adopted JPEG format. However, there is a lack of methods to detect double compression in the HEIF format, which has recently gained popularity since it allows for reduced file size while maintaining image quality. Traditional JPEG-based techniques do not apply to HEIF due to its distinct encoding algorithms. We previously proposed a method to detect double compression in HEIF images based on Farid’s work on coding ghosts in JPEG images. However, this method was limited to scenarios where the quality parameter used for the first encoding was larger than for the second encoding. In this study, we propose a lightweight image classifier to extend the existing model, enabling the identification of double-compressed images without heavily depending on the input image’s quantization history. This extended model outperforms the previous approach and, despite its lightness, demonstrates excellent detection accuracy. Full article

The realization of a BRST cohomology of the 4D topological gauge-affine gravity is established in terms of a superconnection formalism. The identification of fields in the quantized theory occurs directly as is usual in terms of superconnection and its supercurvature components with the double covering of the general affine group $\overline{GA}(4,\mathbb{R})$. Then, by means of an appropriate decomposition of the metalinear double-covering group $\overline{SL}(5,\mathbb{R})$ with respect to the general linear double-covering group $\overline{GL}(4,\mathbb{R})$, one can easily obtain the enlargements of the fields while remaining consistent with the BRST algebra. This leads to the descent equations, allowing us to build the observables of the theory by means of the BRST algebra constructed using a $\overline{sa}(5,\mathbb{R})$ algebra-valued superconnection. In particular, we discuss the construction of topological invariants with torsion. Full article

Generative models have the potential to revolutionize 3D extended reality. A primary obstacle is that augmented and virtual reality need real-time computing. Current state-of-the-art point cloud random generation methods are not fast enough for these applications. We introduce a vector-quantized variational autoencoder model (VQVAE) that can synthesize high-quality point clouds in milliseconds. Unlike previous work in VQVAEs, our model offers a compact sample representation suitable for conditional generation and data exploration with potential applications in rapid prototyping. We achieve this result by combining architectural improvements with an innovative approach for probabilistic random generation. First, we rethink current parallel point cloud autoencoder structures, and we propose several solutions to improve robustness, efficiency and reconstruction quality. Notable contributions in the decoder architecture include an innovative computation layer to process the shape semantic information, an attention mechanism that helps the model focus on different areas and a filter to cover possible sampling errors. Secondly, we introduce a parallel sampling strategy for VQVAE models consisting of a double encoding system, where a variational autoencoder learns how to generate the complex discrete distribution of the VQVAE, not only allowing quick inference but also describing the shape with a few global variables. We compare the proposed decoder and our VQVAE model with established and concurrent work, and we prove, one by one, the validity of the single contributions. Full article