Dianle Zhou - Academia.edu (original) (raw)

Papers by Dianle Zhou

2022 41st Chinese Control Conference (CCC), Jul 25, 2022

This paper presents a new ground-based visual approach for guidance and safe landing of an unmann... more This paper presents a new ground-based visual approach for guidance and safe landing of an unmanned aerial vehicle (UAV) in Global Navigation Satellite System(GNSS)-denied environments. In our previous work, the old system consists of one pan-tilt unit(PTU) with two cameras, whose detection range is limited by the baseline. To achieve long-range detection and cover wide field of regard, we mounted two separate sets of PTU integrated with visible light camera on both sides of the runway instead of our previous assembled stereo vision system. Then, the well-known AdaBoost method was evaluated with regard to detecting and tracking the target. To achieve the relative position between the UAV and landing area, we used triangulation to calculate the 3D coordinates of the UAV. By combining the estimated position in the closed loop control, we obtain the autonomous landing strategy. Finally, we present several real flights in outdoor environments, and compare its accuracy with ground truth provided by GNSS. The results support the validity and accuracy of the presented system.

Network-on-chip (NoC), the main component of multi-core technology, relates closely to interconne... more Network-on-chip (NoC), the main component of multi-core technology, relates closely to interconnection networks for high-performance parallel computers with multiple processors and uses the communication mode of distributed computing system to improve the concurrent of multi-core. Many factors, such as communication latency, area, throughput, power dissipation, etc., are essential to be considered while designing NoC. Due to the great development of integrated circuits, the number of processing units integrated on a chip has increased sharply, which leads to the surge of power density. Power consumption issues have become one of the most important factors of on- chip network design. In order to achieve optimal and concurrent comprehensive performance of high density multi-processor, we proposed a Dragonfly Routing Algorithm on Multi-objects by optimal thresholds (DRAM) through both power and latency optimizations. DRAM was validated efficiently on Dragonfly topology of network-on-chip. The experimental results showed that, compared to DATRA, DRAM can reduce 7.41% on average in minimum power-latency, which is from the value of proposed PL model. Besides, it can also decrease runtime for 2.58% on average when n=4.

Lecture Notes in Computer Science, 2017

The infrared thermal imaging technology has been widely used in the industrial and military field... more The infrared thermal imaging technology has been widely used in the industrial and military fields because of the strong anti-interference ability. One of problem of the infrared camera application is calibration processing, especially, for the long focal length infrared camera. In this paper, we propose an on-line calibration method for general infrared camera where the infrared camera installed on the PAN-Tilt Unit(PTU). The majority advantage of proposal method is no need calibration board. First, infrared image matching algorithm using edge oriented histogram (EOH) descriptor to find correspondence between frames by setting the PTU to variant angles. Then we demonstrate the Pan-Tilt (PT) image matching and calibration algorithm, which is used to calculate the infrared camera intrinsic matrix. The experiments are done on different wavelengths and focal length infrared camera. Infrared calibration board and our proposal method result were compared. The experiment results show that the proposed method is robust and efficient. And we used on-line calibration technique for long distance UAV (Unmanned Aerial Vehicle) detection and localization.

Convolution Neural Networks (CNN) have evolved to be the state-of-art technique for machine learn... more Convolution Neural Networks (CNN) have evolved to be the state-of-art technique for machine learning tasks. However, CNNs bring a significant increase in the computation and parameter storage costs, which makes it difficult to deploy on embedded devices with limited hardware resources and a tight power budget. In recent years, people focus on reducing these overheads by compressing the CNN models, such as pruning weights and pruning filters. Compared with the method of pruning weights, the method of pruning filters does not result in sparse connectivity patterns. And it is conducive to the parallel acceleration on hardware platforms. In this paper, we proposed a new method to judge the importance of filters. In order to make the judgement more accurate, we use the standard deviation to represent the amount of information extracted by the filter. In the process of pruning, the unimportant filters can be removed directly without loss in the test accuracy. We also proposed a multilayer pruning method to avoid setting the pruning rate layer by layer. This holistic pruning method can improve the pruning efficiency. In order to verify the effectiveness of our algorithm, we do experiments with simple network VGG16 and complex networks ResNet18/34. We re-trained the pruned CNNs to compensate the accuracy loss caused by the pruning process. The results showed that our pruning method can reduce inference cost by up to 50% for VGG16 and 35% for ResNet18/34 on CIFAR10 with little accuracy loss.

Industrial Robot-an International Journal, Jun 15, 2015

ABSTRACT Purpose – This study aims to investigate if smartphone sensors can be used in an unmanne... more ABSTRACT Purpose – This study aims to investigate if smartphone sensors can be used in an unmanned aerial vehicle (UAV) localization system. With the development of technology, smartphones have been tentatively used in micro-UAVs due to their lightweight, inexpensiveness and flexibility. In this study, a Samsung Galaxy S3 smartphone is selected as an on-board sensor platform for UAV localization in Global Positioning System (GPS)-denied environments and two main issues are investigated: Are the phone sensors appropriate for UAV localization? If yes, what are the boundary conditions of employing them? Design/methodology/approach – Efficient accuracy estimation methodologies for the phone sensors are proposed without using any expensive instruments. Using these methods, one can estimate his phone sensors accuracy at any time without special instruments. Then, a visual-inertial odometry scheme is introduced to evaluate the phone sensors-based path estimation performance. Findings – Boundary conditions of using smartphone in a UAV navigation system are found. Both indoor and outdoor localization experiments are carried out and experimental results validate the effectiveness of the boundary conditions and the corresponding implemented scheme. Originality/value – With the phone as a payload, UAVs can be further realized in smaller scale at lower cost, which will be used widely in the field of industrial robots.

Proceedings of SPIE, Dec 24, 2013

The Pan-Tilt Unit (PTU) and camera composed a vision system, which can be used in vision measurem... more The Pan-Tilt Unit (PTU) and camera composed a vision system, which can be used in vision measurement. In this paper, a new calibration method for this vision system using differential GPS (global positioning system) is described. The calibration method is an efficient solution for large field of view. The proposed method has advantages in speed, convenience and pertinence. The experimental results show the accuracy is acceptable for the outdoor environment. Keywords—calibration, large feild, Differential GPS, vision system

IEEE Access, 2019

With the refinement of tasks in artificial intelligence, bringing in exponential level increments... more With the refinement of tasks in artificial intelligence, bringing in exponential level increments in computation cost and storage. Therefore, the augment of computation resource for complicated neural networks severely hinders their applications on limited-power devices in recent years. As a result, there is an impending necessity to compress and accelerate the deep networks by special ways. Considering the different peculiarities of weight quantization and sparse regularization, in this paper, we propose a low rank sparse quantization (LRSQ) method to quantize network weights and regularize the corresponding structures at the same time. Our LRSQ can: 1) obtain low-bit quantized networks to reduce memory and computation cost and 2) learn a compact structure from complex convolutional networks for subsequent channel pruning which has significant reduction on FLOPs. In experimental sections, we evaluate the proposed method on several popular models such as VGG-7/16/19 and ResNet-18/34/50, and results show that this method can dramatically reduce parameters and channels of the network with slight inference accuracy loss. Furthermore, we also visualize and analyze the four-dimensional weight tensors, which shows the low rank and groupsparsity structure of it. Finally, we try pruning unimportant channels which are zero-channels in our quantized model, and finding even a little better precision than the standard full-precision network.

HAL (Le Centre pour la Communication Scientifique Directe), Jan 26, 2010

Le projet ANR CSOSG2007 KIVAOU vise à développer un démonstrateur comprenant des outils innovants... more Le projet ANR CSOSG2007 KIVAOU vise à développer un démonstrateur comprenant des outils innovants d'analyse vidéo dédiés à deux problématiques : 1) Un dispositif mobile (valise) d'identification et indexation biométrique faciale portable pour une analyse temps réel vidéo et 2) Une plate-forme d'analyse de vidéos multiples enregistrées lors d'un évènement, utilisant la synchronisation de vidéos, l'extraction de signatures pour les personnes, et la constitution de trajectoires. Le but est de permettre ou de faciliter une analyse a posteriori des données enregistrées en un lieu et pendant une même période à des fins d'investigations. Le projet KIVAOU réunit Sagem Sécurité, EVITECH, FACING-IT, le Ministère de l'Intérieur, l'Institut TELECOM, et ARMINES. Il propose des approches innovantes basées sur la combinaison de briques de base maitrisées par chacun des partenaires: biométrie, analyse d'image, analyse vidéo, suivi, synchronisation, besoins IHM utilisateurs, etc. Il comporte une phase de test auprès des utilisateurs finaux pour valider ou améliorer la pertinence des outils proposés.

The technology of binocular camera matures day by day. Compared with monocular camera, it can obt... more The technology of binocular camera matures day by day. Compared with monocular camera, it can obtain higher resolution images at a lower cost than monocular cameras. However, existing high dynamic range methods based on images acquired by monocular camera, causing the result images to be noisy and blurry. In order to solve the problem, this paper presents a new high dynamic range method based on monochrome-color camera system. We first use the camera system to obtain multiple sets of different exposure monochrome-color image pairs, and then match the same exposure image pair. By using the color propagation methods, we combine the color information from color image with detail information from monochrome image, and obtain multiple sets of different exposures, sharper, low-noise images with more details. And finally get the result through high dynamic imaging and tone mapping. Experiments show that our method is better than the results of the classical method.

Chinese Control Conference, Jul 26, 2013

Hybrid Systems, formed by the combination of thermal and visible sensors, are superior or complem... more Hybrid Systems, formed by the combination of thermal and visible sensors, are superior or complementary to conventional visible-spectrum cameras for many applications. This paper proposes a method for large FOV thermal/visible hybrid binocular vision system calibration based on a new designed calibration pattern. Small circular constant heat reservoir are used as calibration marker for thermal cameras while the visible cameras can obtain the center of the concentric black-white circles as calibration marker with high precision. The proposed method can be used to simultaneously reference both thermal and visible cameras to a global coordinate frame. From the experiments we have obtained the mean re-projection error of calibration in 0.786 pixels at 10 meters. This method is simple and the results are acceptable.

Aerospace, Jan 18, 2023

This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY

Computational Intelligence and Neuroscience, Feb 18, 2020

e increase in sophistication of neural network models in recent years has exponentially expanded ... more e increase in sophistication of neural network models in recent years has exponentially expanded memory consumption and computational cost, thereby hindering their applications on ASIC, FPGA, and other mobile devices. erefore, compressing and accelerating the neural networks are necessary. In this study, we introduce a novel strategy to train low-bit networks with weights and activations quantized by several bits and address two corresponding fundamental issues. One is to approximate activations through low-bit discretization for decreasing network computational cost and dot-product memory. e other is to specify weight quantization and update mechanism for discrete weights to avoid gradient mismatch. With quantized low-bit weights and activations, the costly full-precision operation will be replaced by shift operation. We evaluate the proposed method on common datasets, and results show that this method can dramatically compress the neural network with slight accuracy loss.

International Journal of Advanced Robotic Systems, 2014

This paper presents a framework for tracking a mobile ground target (MGT) using a fixed-wing unma... more This paper presents a framework for tracking a mobile ground target (MGT) using a fixed-wing unmanned aerial vehicle (UAV). Challenges from pure theories to practical applications, including varying illumination, computational limits and a lack of clarity are considered. The procedure consists of four steps, namely: target detection, target localization, states estimation and UAV guidance. Firstly, the MGT in the wild is separated from the background using a Laplacian operator-based method. Next, the MGT is located by performing coordinate transformations with the assumption that the altitude of the ground is invariant and known. Afterwards, a Kalman filter is used to estimate the location and velocity of the MGT. Finally, a modified guidance law is developed to guide the UAV to circle and track the MGT. The performance of our framework is validated by simulations and a number of actual flight tests. The results indicate that the framework is effective and of low computational complexity, and in particular our modified guidance law can reduce the error of the tracking distance by about 75% in specified situations. With the proposed framework, such challenges caused by the actual system can be tackled effectively, and the fixed-wing UAV can track the MGT stably.

This paper employs the Chan-Vese (CV) model into aircraft objective extraction for binocular ster... more This paper employs the Chan-Vese (CV) model into aircraft objective extraction for binocular stereo vision to enable autonomous take-off and landing of unmanned aerial vehicles. Fundamental principles of the CV model and the level set method are summarized as minimizing energy function. Eventually, a flying UAV objective extraction algorithm is proposed and developed by using the CV model. Two sets of UAV landing images are collected for validation. Experimental results show that the proposed algorithm can effectively extract the UAV target even with a complex background. Furthermore, the accuracy of localization is comparable with DGPS and it is better than that BRISK maximal response value algorithm.

Proceedings of SPIE, Feb 12, 2015

In this study, this paper focus on the vision-based autonomous helicopter unmanned aerial vehicle... more In this study, this paper focus on the vision-based autonomous helicopter unmanned aerial vehicle (UAV) landing problems. This paper proposed a multisensory fusion to autonomous landing of an UAV. The systems include an infrared camera, an Ultra-wideband radar that measure distance between UAV and Ground-Based system, an PAN-Tilt Unit (PTU). In order to identify all weather UAV targets, we use infrared cameras. To reduce the complexity of the stereovision or one-cameral calculating the target of three-dimensional coordinates, using the ultra-wideband radar distance module provides visual depth information, real-time Image-PTU tracking UAV and calculate the UAV threedimensional coordinates. Compared to the DGPS, the test results show that the paper is effectiveness and robustness.

Recently, there has been growing interest in developing unmanned aircraft system (UAS) based on v... more Recently, there has been growing interest in developing unmanned aircraft system (UAS) based on visual sensors. During the whole autonomous assignment, the landing procedure is one of the most dangerous and challenging process. For most of unmanned aircraft vehicle, visual sensors are the basic equipment, which are also widely used during the landing maneuver. This paper first presents the main research groups involved in the development of vision-based autonomous landing systems. Then it discusses the detail of each algorithms and systems in different categories. The goal of this paper is to review the state-of-the-art vision-based autonomous landing methods that captures all milestones and seminal works. These algorithms and systems are classified into different categories. Finally, the paper highlights challenges in this research field.

Binocular cameras have gained increasing attention because they can capture high-resolution image... more Binocular cameras have gained increasing attention because they can capture high-resolution images at a lower cost than monocular cameras. However, many existing binocular camera technologies typically require accurate depth estimation. To address this problem, this paper presents a new image enhancement method based on monochromecolored cameras. Our method replaces depth estimation with dense matching of feature points, thereby effectively reducing the computational complexity. After image matching, matrix completion is used to recover the color information of the monochrome image. Consequently, our method produces a high-quality image under the low-light condition. We built real image database for the experiments, and the results reveal that our method exhibits superior performance over existing methods.