Search

There are large number of people with color blindness who have the practical inconveniences in their life and work. The problems they encountered lie in their ignorance of unrecognized color in their own field of vision. Although there are many visual assistance tools provided, most of them work only if the users actually know their trouble recognizing objects, and fixed filters cannot identify shapes of objects perfectly in complex graphics. To conquer the problems, we created an iOS application providing flexible, efficient and real-time image conversion to all color-blind patients. With the dynamic scanning function, the proposed application shows more efficient visual cues and improves patients’ ability to distinguish color combinations.

In the context of advocating the reduction of greenhouse gases, green energy trade has become the focus of attention. At this stage, green energy is relatively dispersed (such as wind power, hydropower, solar power, etc.), and energy generation is real-time. At present, green energy trading based on blockchain has become a better solution. In the transaction itself, the blockchain has better security and traceability, but in terms of statistical information and interval query, because the basic data structure of the blockchain is a key-value data structure, performance problems are prone to occur. A transaction-based static indexing method is proposed, which can effectively improve the efficiency of statistical information and interval query on the chain.

Person re-identification (re-ID) is an important and challenging topic in video surveillance and public security. Re-ID aims to retrieve persons from different cameras. Despite the developments in recent years, re-ID still faces many challenges due to different camera views, changeable person posture, complex background, and occlusion. To exploit more discriminative image similarity descriptor, we propose a novel method in this paper. First, we design a body partition extraction network to extract three body regions with efficient alignment. Second, we propose a multi-stream contribution framework to fuse feature distance with different contributions and generate the final image similarity descriptor. In addition, we combine re-ID and semantic segmentation. A mask feature is introduced to the proposed framework and we design a contribution feedback module to generate contribution coefficients dynamically. Third, in order to improve re-ID performance, we propose a fragment learning method to optimize the contribution feedback module. Fourth and last, we propose a k-distribution re-ranking strategy to further improve performance. Our method achieves competitive results on two popular datasets, CUHK03, and Market1501, with rank-1 accuracy of 93.5% and 85.7%. The proposed re-ranking method achieves 2.3% and 2.8% performance boost. The data demonstrate the effectiveness of the proposed multi-stream contribution framework and the k-distribution re-ranking strategy.

During the process of space travelling and space landing, the spacecraft attitude estimation is the indispensable work for navigation. Since there are not enough satellites for GPS-like localization in space, the computer vision technique is adopted to address the issue. The most crucial task for localization is the extraction of correspondences. In computer vision, optical flow estimation is often used for finding correspondences between images. As the deep neural network being more popular in recent years, FlowNet2 has played a vital role which achieves great success. In this paper, we present PA-FlowNet, an end-to-end pose-auxiliary optical flow network which can use the predicted relative camera pose to improve the performance of optical flow. PA-FlowNet is composed of two sub-networks, the foreground-attention flow network and the pose regression network. The foreground-attention flow network is constructed by FlowNet2 model and modified with the proposed foreground-attention approach. We introduced this approach with the concept of curriculum learning for foreground-background segmentation to avoid backgrounds from resulting in flow prediction error. The pose regression network is used to regress the relative camera pose as an auxiliary for increasing the accuracy of the flow estimation. In addition, to simulate the test environment for spacecraft pose estimation, we construct a 64K moon model and simulate aerial photography with various attitudes to generate Moon64K dataset in this paper. PA-FlowNet significantly outperforms all existing methods on the proposed Moon64K dataset. Furthermore, we also predict the relative camera pose via proposed PA-FlowNet and accomplish the remarkable performance.

In recent years, with the rise of the concept of virtual reality, multidimension-media that contains multiple sensory stimulations such as vision, hearing, and somatosensory have also sprung up. However, existing network streaming platforms are still limited to traditional video and audio streaming, lack the transmission of somatosensory data, and cannot support multidimension-media that requires natural interaction. This paper proposes a multidimension-media streaming system composed of three parties: control center, client platform, and multidimension-media producer. It specifies the login procedures, link procedures, interaction mechanisms, and logout procedures in detail. This architecture adopts the method of separate transmission of audiovisual information and body joint information to solve the problem of the unreturnable somatosensory information in the current streaming system. In this paper, a complete system has been set up to actually verify the flow of producers and client platforms from login, data transmission, stream creation to logout. It proposes an objective function for the system delay and tests the influence of the network cache size on the client platform. The delay can be reduced to less than one second under the best settings. This paper also measured the relationship between the body joints and the network traffic of the control center when a single pair of client-producer were connected and proposed related formulas for reference when building systems in the future.

As the demand for the Internet of Things increases, wireless network protocols applied to the Large-Scale Internet of Things (LS-IoT) is the future development direction. Although 5G technology is a recent development trend, its signal coverage and cost will be key issues. However, the Wi-Fi protocols have many protocol-related modules that are quite mature and relatively inexpensive. And it has the advantages of low latency and high transmission efficiency. We implement Adaptive Tree Topology Algorithm in the Wi-Fi tree network system for Internet of Things data collection and make appropriate design corrections to overcome the influence of environmental factors. In the implementation and testing, it was found that multiple problems caused by environmental factors. First, signal interference in the environment will cause packet loss in the system. Cause the routing mechanism to collapse. Second, the strength of the packet signal is weakened due to obstacles. In addition, the previously designed hop selection mechanism may cause unstable data paths and increase the probability of disconnection. Third, the signal interference affects the transmission of update routing-related packets, which causes the routing data of the node and the gateway to be out of sync. In order to overcome the above-mentioned influence, we designed a Resend Queue retransmission mechanism to effectively solve packet loss, and improve the retransmission time interval to reduce the probability of packet collision due to frequent retransmissions when the interference is high. In addition to the hop selection mechanism, increase the signal strength as another routing metric to ensure the establishment of a network chain with a smaller hopping with stable signal. The maintenance mechanism is designed to update the data in real time and gather it to the gateway to provide more stable routing maintenance and eliminate routing maintenance errors. Furthermore, we implement our algorithm on CC3220MODA. After testing, we offer three objective function for this system. Confirm the feasibility of the network mechanism design in this paper, and provide setting objective functions and setting suggestions for key parameters in the system.

Identity management (IdM) plays a significant role in managing user identities (IDs). However, IdM is challenging to handle the rapidly rising numerous kinds of Web-based applications nowadays. The OpenID 2.0 communication protocol is an improved solution for managing a user's IDs based on the OpenID URL identity. OpenID URL identity is not very much secure in specific Web-based attacks; for instance, session hijacking and phishing attacks often occur. The earlier OpenID-based methods secure OpenID URL identity with single, double, and triple authentication schemes. But Identity Provider (IdP) side is still not secure in Web attacks: if an attacker steals the IdP-side legal user information, then existing OpenID-based security techniques are unreliable. The anticipated OpenID Reverse Authentication Authorizing and Accounting (RAAA) user authentication-based protocol secured OpenID URL identity by providing two beneficial fields Secret Alphanumeric String (SAS) and Special Innovative PIN (SIP) that utilize in testing website both sides in reverse and cost-effective way. In this experiment, IdP and Relying Party (RP), both sides are being used secretly. Therefore, experimental websites also test to check the proposed triple authentication protocol. In this paper, we have compared our RAAA user authentication protocol with already available SSO protocol methods. The tested websites and comparative results represent that the anticipated design protocol is very much secure and reliable solution. The advanced cryptographic Single-Sign-On (SSO) secure protocol reduces the higher-level session hijacking and phishing attacks risk in an OpenID-based environment. We suggest future SSO protocol methods will be needed more in terms of the authorized user's identity authentication in Web-based applications.

Skillful rate-control inside motion estimation (ME) plays a crucial part in real-time mobile video coding because the available transmission bandwidth in mobile video coding applications within handhold devices is time-varying instead of fixed in the video coding standard H.264/AVC or H.265/high efficiency video coding (HEVC). Rate control consists of coding bit allocation and quantization parameter selection, which are difficult to implement in hardware. To overcome this issue, this paper achieves an ME rate control design on the basis of the improved machine learning (ML) scheme, and the auxiliary experimental outcomes reveal the coding effectiveness and feasibility of the proposed method.

Warning the rear approach of objects before opening the car door is essential for safety reasons. Particularly in Taiwan, the traffic accident caused by car door opening were about 20% in 2016 and resulted in a lot of motorcyclists and cyclists being suffered from severe injury. For this, we propose an anti-collision system by using single digital camera and a Field Programmable Gate Array (FPGA) chip. The proposed system first detects approaching objects by motion detection and then derives their speeds to raise an alarm. Here, the warning threshold is set at 16 km/hr to detect an object from 3.8 to 18.8 meters away. Our experimental results indicate that the proposed system is practical for smart car designs with embedded collision detection.

Distance metric learning has been shown to be an effective and efficient method which can lead to significant improvements in classification, clustering, and retrieval. Conventional metric learning methods which apply a global metric to capture the input features and their correlations are unsuitable for heterogeneous data, whose features vary according to different clusters. Thus, a number of local metric learning methods have been proposed to overcome this limitation. Unfortunately, most of them suffer from over-fitting or need the data set to be under a strong assumption, such as Gaussian distribution. To avoid over-fitting caused by learning local metric matrices separately in the previous researches, we propose a novel local metric learning method in which we learn a metric matrix for each pair of instances as linear combinations of basis metric matrices defined on different clusters of the input space without any assumption for data set. We get the parameters of the linear combinations by using the relationship between direct distance and anchor~distance between a pair of data points. Additionally, we design a novel expectation-maximization based optimization method to learn the linear combinations and basis metric matrices. Experimental results show that our method outperforms state-of-the-art methods on various data sets.

The problem of interaction of microwaves with magnetized plasma is complicated and has garnered considerable interest in recent decades. It has been shown that the parameters, such as magnetic field intensity, plasma density, collision frequency, and plasma size, have substantial impacts on microwave propagation in magnetized plasma. The influences of one of the magnetized plasma parameters on the propagation can be studied easily; however, plasma must be considered integrally as it is an electrically neutral medium. In this paper, using a designed magnetized plasma-metal model, we assess how magnetized plasma parameters influence the interaction of microwaves in magnetized plasma in order to reveal a non-monotonic and complex relationship. We numerically investigate the reflection of fixed-frequency and broadband microwaves using the bounded magnetized plasma-metal model. And the optimized plasma parameters, such as plasma frequency, and collision frequency can be given automatically. By optimizing the reflection of fixed-frequency and broadband microwaves, optimized plasma parameters can be obtained. The relationship between microwave reflections and plasma parameters are depicted visually via figures. Finally, we discuss the degree of influence of plasma parameters on the microwave reflection. By parameter optimization, we obtain plasma parameters for plasma device design that meets the requirements of microwave propagation in magnetized plasma. These results may be useful in applications of electromagnetic wave propagation in magnetized plasma.

Today the Web has become an essential part of people’s life. Everyone has the right to access websites, including seniors and disabled people. Government websites contain public information for the whole society, but affect mostly disabled people. In this paper, we have analyzed the accessibility and usability issues of selected top 60 government websites of Pakistan and China. We analyzed all websites’ loading time, page size, mobile friendliness, broken links, HTML5 and ARIA, Web Content Accessibility Guidelines (WCAG) 2.0 (Level A, Level AA, Level AAA) conformance, and contrast errors through online accessible testing tools and human inspections. We designed an interactive evaluation framework for Web accessibility and usability, which is used to evaluate both countries’ government websites. According to the evaluation 22 out of 30 Chinese government websites have less than about 50% of the accessibility issues, while Pakistan in contrast only 12 out of 30 government websites have less than about 50% of accessibility issues. The Chinese government websites are much better than Pakistani government websites with respect to accessibility and usability features. The all-inclusive results indicate that Pakistan government websites have poor conformance levels to meet WCAG standards. We gave some valuable recommendations, which will be helpful for web designers, developers and also web users particularly disabled people performance in the newest Artificial Intelligence (AI) area in a decent way.

Domain adaptation, which is able to leverage the abundant supervision from the source domain and limited supervision in the target domain to construct a model for the data in the target domain, has drawn significant attentions. Most of the existing domain adaptation methods elaborate to map the information derived from the source domain to the target domain for model construction in the target domain. However, such a 'Source' (S) to 'Target' (T) mapping usually involves 'tailoring' the information from the source domain to fit the target domain, which may lose valuable information in the source domain for model construction. Moreover, such a mapping is usually tightly coupled with the model construction, which is more complex than a separate model construction or mapping construction. In this paper, we provide an alternative way for domain adaptation, named T2S. Instead of mapping the 'S' to 'T' and constructing a model in 'T', we inversely map 'T' to 'S' and reuse the model that has been well-trained with abundant information in 'S' for prediction. Such an approach enjoys the abundant information in source domain for model construction and the simplicity of learning mapping separately with limited supervision in target domain. Experiments on both synthetic and real-world data sets indicate the effectiveness of our framework.

CT imaging technology plays an important role in the field of medical diagnosis. Filtered Back-Projection (FBP) algorithm is usually used for image reconstruction. However, there are some noise defects in the reconstructed image, and CT imaging requires higher accuracy of the image. In order to reduce these noise defects, use TT transform instead of Fourier filtering in FBP, and a TT back projection algorithm (TTBP) is proposed. Firstly, The image projection signal is transformed by TT. Then, using the TT transform spectral features of the diagonal elements, the noise signal of the projected signal is processed. Finally, back projection is used to reconstruct the original image. Using Shepp-Logan skull in medical CT imaging as a model and algorithm is implemented in MATLAB. Compared with the simulation results, the obtained image has better control of noise flaws, and the resulting image is clearer and smoother, which has a high application value.