Deep neural networks are frequently used to automate the examination of radiographic images in medical. These approaches may be used to train on huge datasets or extract features from small datasets using pre-trained networks. Due to the lack of large pulmonary tuberculosis datasets, it is possible to diagnose tuberculosis using pre-trained deep convolutional neural networks. Thus, this article aims to detect and diagnose tuberculosis in chest X-rays by combining a pre-trained deep convolutional neural network with a machine learning model. Combined the deep pre-trained DenseNet201 network with the machine learning XGBoost classifier to create a hybrid model for classifying patients as tuberculosis infected or not. The proposed model extracts feature using the pre-trained DenseNet201 neural networks and classify them employing the XGBoost classifier. We performed extensive experiments to assess the performance of the proposed DenseNet201-XGBoost model using tuberculosis chest x-ray images. Comparative study shows that the proposed DenseNet201-XGBoost-based tuberculosis classification model outperforms other competing approaches.
ESA is working with NASA to plan and carry out an international Mars Samples Return (MSR) campaign between 2020 and 2030. A relevant part of the upcoming MSR mission is the Sample Fetch Rover (SFR), tasked to collect sample tubes of Martian soil prepared by Mars2020 rover Perseverance. This work focuses on the localization capabilities of SFR and the potential reuse of the functionalities present in the ExoMars rover. Visual Odometry (VO), a vision-based localization algorithm, is often the main component of the localization process in planetary robotics. The goal of this study is to investigate the possibility of transferring the ExoMars VO solution to a valid SFR implementation, compliant with mission requirements. First, the main differences between the two missions, SFR and ExoMars have been studied, in order to identify the most critical parameters for the VO process. Then, using a testing rover available in the Planetary Robotics Lab (PRL) at European Space TEchnology and research Centre (ESTEC), the effect of the previously identified parameters on the VO performances was evaluated, identifying the most crucial ones and proposing some solutions to face them. This work could lead the way to future studies about the localization for the Sample Fetch Rover and what are the main and most critical factors that would have to be taken into account in order to achieve an accurate and reliable localization system.
Semantic segmentation with image RGB information is significantly useful for intelligent perception of robotics. However, semantic segmentation with only RGB information does not perform well for objects with the same color during grasping manipulation. This paper proposes a new semantic segmentation scheme based on the fusion of RGB and heights transformed from depth information, which is not simple fusion of RGB-D method. It modifies the height information so that different objects of the same color can be distinguished in height. It outperforms the classical RGB segmentation scheme at improving speed and 7.42% higher at the final performance of semantic segmentation of manipulator grasping scene (contains objects with the same color). Because of the need of RGB-D information, this paper proposes a method of self-collecting and self-labeling data of manipulator grasping scene, which reduces the cost of manpower by making full use of the highly automated equipment and the characteristics of the scene.
Some of the most appealing areas for future planetary surface exploration lie in rough, uneven terrains, such as craters and cold traps, which are currently inaccessible by state-of-the-art robotic systems [1]. To provide modularity for access and in-situ sampling within such extreme environments, Jet Propulsion Laboratory (JPL) and California Institute of Technology have collaborated to develop the DuAxel rover system, a modular robot composed of two Axel rappelling vehicles docked to a central module into a four-wheeled configuration, suited for driving long distances due to articulated passivesteering capabilities [2]. Inspired by Carnegie Mellon’s Zoë rover [3], a 3D kinematic control strategy, leveraging a novel, DuAxel-centric model, has been developed to enable precise trajectory following over rough, flat terrain with presence of obstacles.
Segmentation is a key process of 2-D barcode identification. In this paper we propose two 2-D barcode image segmentation algorithms under complex background. The first algorithm is an expansion of multi-scale morphology reconstruction and it can acquire a good segmentation result. However, this algorithm is not suitable for fast real-time image processing due to large computation. To fix the shortcoming of the first algorithm we put forward a fast real-time image segmentation algorithm. The proposed approach is applied in experiments on 2D barcodes with complicated background. In experiments, the second proposed method can maintain image segmentation accuracy while significantly reducing the time consumption.
To solve the adverse effects brought by resource node transfering the using right to local task and the difficult problem of resource load balancing, a two-phase pricing strategy based on QoS constraints is proposed in this paper. On the premise of guaranteeing the benefits of the resource provider in the cost price, this strategy balances the load of the resource provider by the profit price. The theoretical analysis proves the effectiveness of the pricing strategy, and the algorithm of the pricing strategy is designed in this paper. Resources node information in the real distributed systems is used as the performance parameters of experimental node in the simulation experiments, and the performance of the pricing strategy is tested in a large-scale grid mission. Experimental results show that, compared with the traditional pricing strategies, the two-phase pricing strategy based on QoS constraints has vastly superior performance on the benefits of the resource provider and the balance of resource utilization.
Image processing and pattern recognition have been applied to the image acquisition, image enhancement, extraction of classical structure characteristics, binary image processing, and image segmentation of synthetic minerals. They have also been used for the classification, recognition, testing, and survey of the classical structure characteristics, and with the information obtained, a complete image description can be created, from a two-dimensional image to concrete data and quantitative information on the characteristic structure. The results show that image processing and pattern recognition can be successfully applied to the analysis of classical characteristics (such as pore size, number, form, distribution, etc.) of synthetic mineral microstructures, including pore defect and characteristic mineral phase. Information on synthetic mineral microstructure characteristics can be obtained visually, qualitative, quantitatively, and automatically, producing a new research method and testing tool for pore defect research of synthetic minerals.
Focused on the issue that the mobile terminal can not remote control each other, a remote control system between Android platforms was proposed. The system is a typical C/S mode, follow the specification of Android application, to achieve the remote control between two Android system developed by Java language. Firstly, it analyzes the key technologies of the system, like the architecture of Android system, RFB protocol and Java Socket system; Secondly, it builds the structure model of the system, system framework, system module hierarchy and system processes; then, it analyzes in detail the function of the system layers, each module function and system processes; Finally, it realizes the remote control between Android platform through the testing on a mobile device.
Traditional classroom teaching is difficult to satisfy students' personalized learning. The main work of the thesis constructs knowledge point model which using relational model and constructs student model according to the students' interest characteristics, it uses improved k-means algorithm to analyze students' learning interest characteristics to dig out the interests of different students, and provide the advice for teaching strategies, so as to meet the demand of students' personalized learning. Ultimately we improve the teaching quality of teachers and students to learn the effect of teaching optimization purposes.
Using fish behavior to reflect changes of water quality and aim at water quality warning is currently a popular topic of environmental monitoring. The fish tracking technology with occlusion is the basis for the analysis of fish school behavior, and only tracking each individual fish accurately in the shoal can get the exact quantitative analysis of their feature. In this paper, we design the data acquisition platform by using a plane mirror and a monocular camera, and present a fish tracking with occlusion method by using the three-dimensional information. According to the principle of plane mirror imaging and the camera imaging, we find the relationship between the target in mirror-view and the one in the direct-view. Based on the relationship and underwater imaging process, we calculate the 3D coordinates-for both real fish and imaginary fish under circumstance of occlusion. Finally, we use the 3D coordinate to achieve the target occlusion tracking for fish school. The experimental results show that this approach can obtain precise tracking of three fish and get the 3D coordinate accurately even they are in occlusion.
With the rapid development of computer technology, Software-Defined Radio has become more and more popular. We can now transfer the communication system from hardware to software, which makes it more flexible and helps us virtualize the communication system. In this paper, we creatively introduce a method to guarantee hard real-time in user-space on Linux. Unlike the usual methods, Our method(real time with RDTSC) provides real-time feature by soft ware clock with RDTSC and isolating CPUs instead of interrupts from the hardware. Our method is highly lightweight and give the SDR systems more flexibility when compared to the usual methods with interrupts. The experiment and validation results show that real time with RDTSC has a significantperformance and can be used in physical communication systems.
Symbol timing synchronization is an important component in communication systems. The quality and reliability of wireless network communication will improve with accurate symbol timing recovery. In this paper, we present a symbol timing synchronization method by interpolating the samples at the output of the matched filter to compute the desired time instants. Since interpolation jitter is the major disadvantage to this method, the approach to remove the interpolation jitter has also been proposed. Finally, hardware implementation details of the method to realize symbol timing synchronization for an MQAM system have been provided. The simulation and experiment results show that the design meets the timing requirements of the system.
Target advertising is ubiquitous wherever we get online with any device. There is nothing unusual to click in and purchase for our potential desire. However, it may reveal the history we have browsed while sharing the device to others. Besides, when we alternate use the different devices, the target advertising is unaware of our switching, which leads to the miss of useful target advertising.This paper proposes a system called Prouter to preserve users' profile from being leaked through targeting advertising. Prouter consists of two modules, namely smart router and secure plug-in. The smart router serves as a pivot of the Prouter, processing the data and distinguishing each member from current users. It will vouch for the real identity of each member. The secure plug-in aims to replacing and blocking the advertisement before loading the page. Evaluation depends on the feedback we displayed in real condition, which shows that Prouter can distinguish current user from the family members.
For intelligent vehicle systems, lane detection is still a challenging task because it must cope with various road environments. In this paper, we propose a reliable method with Gabor filters. The proposed approach consists of two step. In the first step, the vanishing-point locations is estimated by the texture feature based method. The key attributes of this method consist of the dominant texture orientations, the vanishing-point candidates' vote weighting based on its dominant orientation, and a soft voting scheme. In the second step, the road lane ahead of the vehicle is detected by edge detection, and the vanishing-point is used to constrain a search for the lane mark. This method is insensitive to variations of road condition and illuminations.
Current surveillance system in some specific places like bank, hospital or train station are not efficient enough for city manager or policeman. Huge amount of manpower are needed to make full use of these surveillance data in crime research. In this paper, a novel highly automated surveillance system are proposed. The system would get the identification information and human position indirectly by finding the cellphone they carry on. By running the VBTS (Virtual Base Transceiver Station), IMSI (International Mobile Subscriber Identification Number) of every mobile phone would be collected. VBTS is also able to make cellphones continuously transmit signal. A small sensor is designed to collect the RSS (Receive Signal Strength) and get cellphones' location by triangulation algorithm.
Near infrared spectroscopy with support vector machine (NIR-SVM) to predict the crude protein (CP) in Alfalfa samples. The R-2 of the predicted CP versus the experimental CP of the training data set is 0.983. The R-2 of the independent test Data Set is 0.9823. The result suggested that it is feasible to rapidly determine the CP of Alfalfa by NIR data based on SVR.
In order to solve the problem that the rear closing target were collided because drivers and passengers opened door abruptly, a door open warning (DOW) system was presented. The proposed DOW system includes system and radar platform structure and key technology. Line Frequency Modulation Continuous Wave (LFMCW) radar sensor was used to really detect the closing targets in the rear of the vehicle. In the same time, cell maximum and minimum average-CFAR (CMMA-CFAR) algorithm was proposed to maintain higher detection rate by adjustment threshold in time based on the noise intensity. The DOW system is implemented on a DSP-based embedded platform. System was calibrated and tested on the Chery Arrizo7 car. For three representative closing targets: bicycles, motorcycles and motor vehicles, the early average warning rates were up to 98.00%, false alarm rate was down to 3.50%. The experimental results show that the proposed DOWS can really detect the moving targets which were into the behind warning area of the vehicle and give collision warning to driver and passenger effectively in various daytime and nighttime road environments.
The Traveling Salesman Problem (TSP) belongs to the class of NP-hard optimization problems. Its solving procedure is complicated, especially for large scale problems. In order to solve the large scale TSPs efficiently, this paper presents a bilevel genetic algorithm with clustering (BLGAC). BLGA-C uses a clustering method to divide a large scale TSP into several subproblems, each subproblem corresponds to a cluster. K-Means clustering method is adopted in this paper. In the lower level, a genetic algorithm is used to find the shortest hamiltonian cycle for each cluster. All these clusters can be handled parallelly. Then, we need to select two nearest vertices between two clusters, and determine which edges will be deleted from the shortest hamiltonian cycle for each cluster, and which edges will be linked for combining two adjacent clusters into one. Repeat this procedure until all clusters are joined into one whole tour. Different combing sequences among clusters will result in different travelling tours, searching for the shortest is our purpose. Therefore, in the higher level, a modified genetic algorithm is designed for integral optimization with the objective of shortest the whole traveling tour. At last in this paper, we trial run a set of experiments on benchmark instances for testing the performance of the proposed BLGAC. Experimental results demonstrate its effective and efficient performance.
