arXiv subjects

Man-On Pun

Publications and source records attributed to Man-On Pun.

33 records · Page 2Linked to original sources

arXiv2024

DF4LCZ: A SAM-Empowered Data Fusion Framework for Scene-Level Local Climate Zone Classification

Recent advancements in remote sensing (RS) technologies have shown their potential in accurately classifying local climate zones (LCZs). However, traditional scene-level methods using convolutional neural networks (CNNs) often struggle to integrate prior knowledge of ground objects effectively. Moreover, commonly utilized data sources like Sentinel-2 encounter difficulties in capturing detailed ground object information. To tackle these challenges, we propose a data fusion method that integrates ground object priors extracted from high-resolution Google imagery with Sentinel-2 multispectral imagery. The proposed method introduces a novel Dual-stream Fusion framework for LCZ classification (DF4LCZ), integrating instance-based location features from Google imagery with the scene-level spatial-spectral features extracted from Sentinel-2 imagery. The framework incorporates a Graph Convolutional Network (GCN) module empowered by the Segment Anything Model (SAM) to enhance feature extraction from Google imagery. Simultaneously, the framework employs a 3D-CNN architecture to learn the spectral-spatial features of Sentinel-2 imagery. Experiments are conducted on a multi-source remote sensing image dataset specifically designed for LCZ classification, validating the effectiveness of the proposed DF4LCZ. The related code and dataset are available at https://github.com/ctrlovefly/DF4LCZ.

cs.CV↗

arXiv2024

GCD-DDPM: A Generative Change Detection Model Based on Difference-Feature Guided DDPM

Deep learning (DL)-based methods have recently shown great promise in bitemporal change detection (CD). Existing discriminative methods based on Convolutional Neural Networks (CNNs) and Transformers rely on discriminative representation learning for change recognition while struggling with exploring local and long-range contextual dependencies. As a result, it is still challenging to obtain fine-grained and robust CD maps in diverse ground scenes. To cope with this challenge, this work proposes a generative change detection model called GCD-DDPM to directly generate CD maps by exploiting the Denoising Diffusion Probabilistic Model (DDPM), instead of classifying each pixel into changed or unchanged categories. Furthermore, the Difference Conditional Encoder (DCE), is designed to guide the generation of CD maps by exploiting multi-level difference features. Leveraging the variational inference (VI) procedure, GCD-DDPM can adaptively re-calibrate the CD results through an iterative inference process, while accurately distinguishing subtle and irregular changes in diverse scenes. Finally, a Noise Suppression-based Semantic Enhancer (NSSE) is specifically designed to mitigate noise in the current step's change-aware feature representations from the CD Encoder. This refinement, serving as an attention map, can guide subsequent iterations while enhancing CD accuracy. Extensive experiments on four high-resolution CD datasets confirm the superior performance of the proposed GCD-DDPM. The code for this work will be available at https://github.com/udrs/GCD.

cs.CV↗

arXiv2024

Diffusion Enhancement for Cloud Removal in Ultra-Resolution Remote Sensing Imagery

The presence of cloud layers severely compromises the quality and effectiveness of optical remote sensing (RS) images. However, existing deep-learning (DL)-based Cloud Removal (CR) techniques encounter difficulties in accurately reconstructing the original visual authenticity and detailed semantic content of the images. To tackle this challenge, this work proposes to encompass enhancements at the data and methodology fronts. On the data side, an ultra-resolution benchmark named CUHK Cloud Removal (CUHK-CR) of 0.5m spatial resolution is established. This benchmark incorporates rich detailed textures and diverse cloud coverage, serving as a robust foundation for designing and assessing CR models. From the methodology perspective, a novel diffusion-based framework for CR called Diffusion Enhancement (DE) is proposed to perform progressive texture detail recovery, which mitigates the training difficulty with improved inference accuracy. Additionally, a Weight Allocation (WA) network is developed to dynamically adjust the weights for feature fusion, thereby further improving performance, particularly in the context of ultra-resolution image generation. Furthermore, a coarse-to-fine training strategy is applied to effectively expedite training convergence while reducing the computational complexity required to handle ultra-resolution images. Extensive experiments on the newly established CUHK-CR and existing datasets such as RICE confirm that the proposed DE framework outperforms existing DL-based methods in terms of both perceptual quality and signal fidelity.

eess.IV↗

arXiv2023

SAM-Assisted Remote Sensing Imagery Semantic Segmentation with Object and Boundary Constraints

Semantic segmentation of remote sensing imagery plays a pivotal role in extracting precise information for diverse down-stream applications. Recent development of the Segment Anything Model (SAM), an advanced general-purpose segmentation model, has revolutionized this field, presenting new avenues for accurate and efficient segmentation. However, SAM is limited to generating segmentation results without class information. Consequently, the utilization of such a powerful general vision model for semantic segmentation in remote sensing images has become a focal point of research. In this paper, we present a streamlined framework aimed at leveraging the raw output of SAM by exploiting two novel concepts called SAM-Generated Object (SGO) and SAM-Generated Boundary (SGB). More specifically, we propose a novel object loss and further introduce a boundary loss as augmentative components to aid in model optimization in a general semantic segmentation framework. Taking into account the content characteristics of SGO, we introduce the concept of object consistency to leverage segmented regions lacking semantic information. By imposing constraints on the consistency of predicted values within objects, the object loss aims to enhance semantic segmentation performance. Furthermore, the boundary loss capitalizes on the distinctive features of SGB by directing the model's attention to the boundary information of the object. Experimental results on two well-known datasets, namely ISPRS Vaihingen and LoveDA Urban, demonstrate the effectiveness of our proposed method. The source code for this work will be accessible at https://github.com/sstary/SSRS.

cs.CV↗

arXiv2023

UniTSA: A Universal Reinforcement Learning Framework for V2X Traffic Signal Control

Traffic congestion is a persistent problem in urban areas, which calls for the development of effective traffic signal control (TSC) systems. While existing Reinforcement Learning (RL)-based methods have shown promising performance in optimizing TSC, it is challenging to generalize these methods across intersections of different structures. In this work, a universal RL-based TSC framework is proposed for Vehicle-to-Everything (V2X) environments. The proposed framework introduces a novel agent design that incorporates a junction matrix to characterize intersection states, making the proposed model applicable to diverse intersections. To equip the proposed RL-based framework with enhanced capability of handling various intersection structures, novel traffic state augmentation methods are tailor-made for signal light control systems. Finally, extensive experimental results derived from multiple intersection configurations confirm the effectiveness of the proposed framework. The source code in this work is available at https://github.com/wmn7/Universal_Light

eess.SY↗

arXiv2023

ADLight: A Universal Approach of Traffic Signal Control with Augmented Data Using Reinforcement Learning

Traffic signal control has the potential to reduce congestion in dynamic networks. Recent studies show that traffic signal control with reinforcement learning (RL) methods can significantly reduce the average waiting time. However, a shortcoming of existing methods is that they require model retraining for new intersections with different structures. In this paper, we propose a novel reinforcement learning approach with augmented data (ADLight) to train a universal model for intersections with different structures. We propose a new agent design incorporating features on movements and actions with set current phase duration to allow the generalized model to have the same structure for different intersections. A new data augmentation method named \textit{movement shuffle} is developed to improve the generalization performance. We also test the universal model with new intersections in Simulation of Urban MObility (SUMO). The results show that the performance of our approach is close to the models trained in a single environment directly (only a 5% loss of average waiting time), and we can reduce more than 80% of training time, which saves a lot of computational resources in scalable operations of traffic lights.

eess.SY↗

arXiv2017

VideoSet: A Large-Scale Compressed Video Quality Dataset Based on JND Measurement

A new methodology to measure coded image/video quality using the just-noticeable-difference (JND) idea was proposed. Several small JND-based image/video quality datasets were released by the Media Communications Lab at the University of Southern California. In this work, we present an effort to build a large-scale JND-based coded video quality dataset. The dataset consists of 220 5-second sequences in four resolutions (i.e., $1920 \times 1080$, $1280 \times 720$, $960 \times 540$ and $640 \times 360$). For each of the 880 video clips, we encode it using the H.264 codec with $QP=1, \cdots, 51$ and measure the first three JND points with 30+ subjects. The dataset is called the "VideoSet", which is an acronym for "Video Subject Evaluation Test (SET)". This work describes the subjective test procedure, detection and removal of outlying measured data, and the properties of collected JND data. Finally, the significance and implications of the VideoSet to future video coding research and standardization efforts are pointed out. All source/coded video clips as well as measured JND data included in the VideoSet are available to the public in the IEEE DataPort.

cs.MM↗

arXiv2009

Distributed Opportunistic Scheduling With Two-Level Probing

Distributed opportunistic scheduling (DOS) is studied for wireless ad-hoc networks in which many links contend for the channel using random access before data transmissions. Simply put, DOS involves a process of joint channel probing and distributed scheduling for ad-hoc (peer-to-peer) communications. Since, in practice, link conditions are estimated with noisy observations, the transmission rate has to be backed off from the estimated rate to avoid transmission outages. Then, a natural question to ask is whether it is worthwhile for the link with successful contention to perform further channel probing to mitigate estimation errors, at the cost of additional probing. Thus motivated, this work investigates DOS with two-level channel probing by optimizing the tradeoff between the throughput gain from more accurate rate estimation and the resulting additional delay. Capitalizing on optimal stopping theory with incomplete information, we show that the optimal scheduling policy is threshold-based and is characterized by either one or two thresholds, depending on network settings. Necessary and sufficient conditions for both cases are rigorously established. In particular, our analysis reveals that performing second-level channel probing is optimal when the first-level estimated channel condition falls in between the two thresholds. Numerical results are provided to illustrate the effectiveness of the proposed DOS with two-level channel probing. We also extend our study to the case with limited feedback, where the feedback from the receiver to its transmitter takes the form of (0,1,e).

cs.IT↗

arXiv2009

Reduced-Feedback Opportunistic Scheduling and Beamforming with GMD for MIMO-OFDMA

Opportunistic scheduling and beamforming schemes have been proposed previously by the authors for reduced-feedback MIMO-OFDMA downlink systems where the MIMO channel of each subcarrier is decomposed into layered spatial subchannels. It has been demonstrated that significant feedback reduction can be achieved by returning information about only one beamforming matrix (BFM) for all subcarriers from each MT, compared to one BFM for each subcarrier in the conventional schemes. However, since the previously proposed channel decomposition was derived based on singular value decomposition, the resulting system performance is impaired by the subchannels associated with the smallest singular values. To circumvent this obstacle, this work proposes improved opportunistic scheduling and beamforming schemes based on geometric mean decomposition-based channel decomposition. In addition to the inherent advantage in reduced feedback, the proposed schemes can achieve improved system performance by decomposing the MIMO channels into spatial subchannels with more evenly distributed channel gains. Numerical results confirm the effectiveness of the proposed opportunistic scheduling and beamforming schemes.

cs.IT↗

arXiv2008

Opportunistic Collaborative Beamforming with One-Bit Feedback

An energy-efficient opportunistic collaborative beamformer with one-bit feedback is proposed for ad hoc sensor networks over Rayleigh fading channels. In contrast to conventional collaborative beamforming schemes in which each source node uses channel state information to correct its local carrier offset and channel phase, the proposed beamforming scheme opportunistically selects a subset of source nodes whose received signals combine in a quasi-coherent manner at the intended receiver. No local phase-precompensation is performed by the nodes in the opportunistic collaborative beamformer. As a result, each node requires only one-bit of feedback from the destination in order to determine if it should or shouldn't participate in the collaborative beamformer. Theoretical analysis shows that the received signal power obtained with the proposed beamforming scheme scales linearly with the number of available source nodes. Since the the optimal node selection rule requires an exhaustive search over all possible subsets of source nodes, two low-complexity selection algorithms are developed. Simulation results confirm the effectiveness of opportunistic collaborative beamforming with the low-complexity selection algorithms.

cs.IT↗

arXiv2008

Opportunistic Scheduling and Beamforming for MIMO-OFDMA Downlink Systems with Reduced Feedback

Opportunistic scheduling and beamforming schemes with reduced feedback are proposed for MIMO-OFDMA downlink systems. Unlike the conventional beamforming schemes in which beamforming is implemented solely by the base station (BS) in a per-subcarrier fashion, the proposed schemes take advantages of a novel channel decomposition technique to perform beamforming jointly by the BS and the mobile terminal (MT). The resulting beamforming schemes allow the BS to employ only {\em one} beamforming matrix (BFM) to form beams for {\em all} subcarriers while each MT completes the beamforming task for each subcarrier locally. Consequently, for a MIMO-OFDMA system with $Q$ subcarriers, the proposed opportunistic scheduling and beamforming schemes require only one BFM index and $Q$ supportable throughputs to be returned from each MT to the BS, in contrast to $Q$ BFM indices and $Q$ supportable throughputs required by the conventional schemes. The advantage of the proposed schemes becomes more evident when a further feedback reduction is achieved by grouping adjacent subcarriers into exclusive clusters and returning only cluster information from each MT. Theoretical analysis and computer simulation confirm the effectiveness of the proposed reduced-feedback schemes.

cs.IT↗

arXiv2008

Distributed Opportunistic Scheduling for MIMO Ad-Hoc Networks

Distributed opportunistic scheduling (DOS) protocols are proposed for multiple-input multiple-output (MIMO) ad-hoc networks with contention-based medium access. The proposed scheduling protocols distinguish themselves from other existing works by their explicit design for system throughput improvement through exploiting spatial multiplexing and diversity in a {\em distributed} manner. As a result, multiple links can be scheduled to simultaneously transmit over the spatial channels formed by transmit/receiver antennas. Taking into account the tradeoff between feedback requirements and system throughput, we propose and compare protocols with different levels of feedback information. Furthermore, in contrast to the conventional random access protocols that ignore the physical channel conditions of contending links, the proposed protocols implement a pure threshold policy derived from optimal stopping theory, i.e. only links with threshold-exceeding channel conditions are allowed for data transmission. Simulation results confirm that the proposed protocols can achieve impressive throughput performance by exploiting spatial multiplexing and diversity.

cs.IT↗

arXiv2008

Distributed Opportunistic Scheduling For Ad-Hoc Communications Under Noisy Channel Estimation

Distributed opportunistic scheduling is studied for wireless ad-hoc networks, where many links contend for one channel using random access. In such networks, distributed opportunistic scheduling (DOS) involves a process of joint channel probing and distributed scheduling. It has been shown that under perfect channel estimation, the optimal DOS for maximizing the network throughput is a pure threshold policy. In this paper, this formalism is generalized to explore DOS under noisy channel estimation, where the transmission rate needs to be backed off from the estimated rate to reduce the outage. It is shown that the optimal scheduling policy remains to be threshold-based, and that the rate threshold turns out to be a function of the variance of the estimation error and be a functional of the backoff rate function. Since the optimal backoff rate is intractable, a suboptimal linear backoff scheme that backs off the estimated signal-to-noise ratio (SNR) and hence the rate is proposed. The corresponding optimal backoff ratio and rate threshold can be obtained via an iterative algorithm. Finally, simulation results are provided to illustrate the tradeoff caused by increasing training time to improve channel estimation at the cost of probing efficiency.

cs.IT↗

arXiv2008

SINR Analysis of Opportunistic MIMO-SDMA Downlink Systems with Linear Combining

Opportunistic scheduling (OS) schemes have been proposed previously by the authors for multiuser MIMO-SDMA downlink systems with linear combining. In particular, it has been demonstrated that significant performance improvement can be achieved by incorporating low-complexity linear combining techniques into the design of OS schemes for MIMO-SDMA. However, this previous analysis was performed based on the effective signal-to-interference ratio (SIR), assuming an interference-limited scenario, which is typically a valid assumption in SDMA-based systems. It was shown that the limiting distribution of the effective SIR is of the Frechet type. Surprisingly, the corresponding scaling laws were found to follow $ε\log K$ with $0<ε<1$, rather than the conventional $\log\log K$ form. Inspired by this difference between the scaling law forms, in this paper a systematic approach is developed to derive asymptotic throughput and scaling laws based on signal-to-interference-noise ratio (SINR) by utilizing extreme value theory. The convergence of the limiting distribution of the effective SINR to the Gumbel type is established. The resulting scaling law is found to be governed by the conventional $\log\log K$ form. These novel results are validated by simulation results. The comparison of SIR and SINR-based analysis suggests that the SIR-based analysis is more computationally efficient for SDMA-based systems and it captures the asymptotic system performance with higher fidelity.

cs.IT↗

arXiv2007

Opportunistic Scheduling and Beamforming for MIMO-SDMA Downlink Systems with Linear Combining

Opportunistic scheduling and beamforming schemes are proposed for multiuser MIMO-SDMA downlink systems with linear combining in this work. Signals received from all antennas of each mobile terminal (MT) are linearly combined to improve the {\em effective} signal-to-noise-interference ratios (SINRs). By exploiting limited feedback on the effective SINRs, the base station (BS) schedules simultaneous data transmission on multiple beams to the MTs with the largest effective SINRs. Utilizing the extreme value theory, we derive the asymptotic system throughputs and scaling laws for the proposed scheduling and beamforming schemes with different linear combining techniques. Computer simulations confirm that the proposed schemes can substantially improve the system throughput.

cs.IT↗