Search

Air-writing is a modern human–computer interaction technology that allows participants to write words or letters with finger or hand movements in free space in a simple and intuitive manner. Air-writing recognition is a particular case of gesture recognition in which gestures can be matched to write characters and digits in the air. Air-written characters show extensive variations depending on the various writing styles of participants and their speed of articulation, which presents quite a difficult task for effective character recognition. In order to solve these difficulties, this current work proposes an air-writing system using a web camera. The proposed system consists of two parts: alphabetic recognition and digit recognition. In order to assess our proposed system, two character datasets were used: an alphabetic dataset and a numeric dataset. We collected samples from 17 participants and asked each participant to write alphabetic characters (A to Z) and numeric digits (0 to 9) about 5–10 times. At the same time, we recorded the position of the fingertips using MediaPipe. As a result, we collected 3166 samples for the alphabetic dataset and 1212 samples for the digit dataset. First, we preprocessed the dataset and then created two datasets: image data and padding sequential data. The image data were fed into the convolution neural networks (CNN) model, whereas the sequential data were fed into bidirectional long short-term memory (BiLSTM). After that, we combined these two models and trained again with 5-fold cross-validation in order to increase the character recognition accuracy. In this work, this combined model is referred to as a hybrid deep learning model. Finally, the experimental results showed that our proposed system achieved an alphabet recognition accuracy of 99.3% and a digit recognition accuracy of 99.5%. We also validated our proposed system using another publicly available 6DMG dataset. Our proposed system provided better recognition accuracy compared to the existing system.

Rear view cameras which help to park or to drive backward has been distributed through after-market, but it is inconvenient because they do not provide views of left, right and front sides. Around view monitoring(AVM) systems which can monitor around vehicle at a glance have been developed and equipped by vehicle vendor but systematic studies on these systems is lack. While the AVM system which equipped on Infiniti cars of Nissan is adequate to monitor around vehicle at a glance, it has disadvantages that additional cabling because of using analog cables is required and image quality is lowered due to EMI/EMC intervention. The around view monitoring system implemented in this paper has advantages that there are no EMI/EMC problems because of using optical network and that cabling is simple because of using plug-and-play ways. Additionally, an advantage of MOST150 network is that camera nodes and display node can be easily installed in the form of plug-and-play.

Recently certain driver convenience systems including ADAS (Advanced Driver Assistance System) such as navigation and GPS, and driver’s infotainment devices such as DVD players have been produced. AVM systems of these driver convenience systems support safety driving by providing situations around the vehicle to a driver. A driver can't fully monitor the situation around the vehicle at a glance using only rearview mirrors, a room mirror, and front and rear cameras, and using of only these devices causes blind spot which hinders driver’s safety. Blind spot is cause of personal injury and vehicle damage when parking vehicle, passing on a narrow street and turning left or right. Therefore studies on AVM systems have actively been progressing which provides surroundings of vehicles for drivers. Though existing AVM systems combines video images from four cameras and provides a driver with a full vision around the vehicle with no blind spots, the system has limitations that its monitoring area is narrow and fixed. Thus they are suitable for parking but are not suitable for monitoring situations around vehicle in various purposes. Therefore, in this paper, we will design and implement AVM system with adjustable around view to be shown that the proposed system provides multiple scopes of around views.

For a variety of vision systems such as car omni-directional surveillance systems and robot vision systems, many cameras have been equipped and used. In order to detect corners of grid pattern in AVM(Around View Monitoring) systems, after the non-linear radial distortion image obtained from wide-angle camera is corrected, corners of grids of the distortion corrected image must be detected. Though there are transformations such as Sub-Pixel and Hough transformation as corner detection methods for AVM systems, it is difficult to achieve automatic detection by Sub-Pixel and accuracy by Hough transformation. Therefore, we showed that the automatic detection proposed in this paper, which detects corners accurately from the distortion corrected image could be applied for AVM systems, by designing and implementing it, and evaluating its performance.

In recent, the researches on driver assistance systems have been actively performed with development of vehicle industry. AVM(Around View Monitoring) Systems, a part of these systems, have been researching and developing. Existing AVM systems have been developed in the forms of embedded systems or a SoC (System on Chip) to provide view around vehicle in real time. However, if Car PC is equipped with in vehicle, AVM can be developed using only software without additional cost. In this study, we implemented AVM system which provides location information by adding the informations such as latitude, longitude and speed to functions of "Car PC" based AVM system. If storing function is added to the AVM system implemented in this study which provides location information, the system with storing function can be used as AVM black box system.

Driver's view has blind spot of automobile surroundings due to physical components of automobile architecture. Obstacles on blind spot are the cause of car destruction and car accidents. Cars which produced in recent have obstacle detection sensors and rear view cameras which provide information of obstacles on the blind sopt, and have also AVM(Around View Monitoring) which provides automobile surroundings for driver's safe driving. During a low-speed travel while parking or moving in a narrow street, a driver get help for safe driving by taking information of automobile surroundings using the above-mentioned devices. In this paper, we present a design and implementation of a 4-sided monitoring (4SM) system, which helps a driver see an integrated view of a vehicle’s perimeter at a glance, using a car PC connected to four cameras installed on the front, rear, left, and right sides.

According to increase of the number of electronic devices for user comfort and safety, electronic systems are being constructed by using network to effectively control the devices. In this paper, we design and implement the MOST network diagnostic system by using wireless LAN to diagnose MOST network and the devices connected by the network and to effectively manage them. MOST, which is a vehicle multimedia network standard for the next generation, is being used for construction of car infotainment systems, and enables to develop various devices which are required for them. The wireless diagnostic system implemented in this paper enables to self-diagnose MOST network and also to check errors by diagnosing status of MOST electronic devices. We also can check and manage status of a in-vehicle MOST network system using mobile devices based on wireless LAN.

While MOST is an in-vehicle network which transfers concurrently synchronous data, asynchronous data and control data, it provides high bandwidth synchronous section which can transfer video and audio without buffering. To transfer real time data using synchronous section, connections between source node and sink node, and channel allocation for connections are required. In this paper, we proposed synchronous data transfer method and channel allocation method by constructing MOST network after designing and implementing channel allocation module for using synchronous data section.

In this paper, the magnetic sensing system is designed and implemented for the safe security in internet commerce system. When the payment is required inthe internet commerce system, the magnetic sensing system will get the information from a credit card without keyboard input and then encrypt and transmit the information to server. The credit card sensing system, which is proposed in this paper, is safe from keyboard hacking because it encrypts card information immediately in its internal chip and sends the information to host system. For the protection of information, the magnetic sensing system is basically based on a synchronous stream cipher cryptosystem which is related to a group of matrices. The size of matrices and the bits of keys for the best performances are determined for various cases. It is shown that for credit card payments. matrices of size 2 have good performance even at most 128bits keys with the consideration of inverse matrices. For authentication of general-purpose data, the magnetic sensing system needs more than 1.5KB data and in this case, the optimum size of matrices is 2 or 3 at more 256bits keys with consideration of inverse matrices.

The automotive electronic systems for effective driving include a navigation reflecting real-time traffic information, a blind spot camera for accident prevention, a black box for recording driving image prior and after accidents, an around view monitoring system enabling all the sides to be seen, etc. The existing around view monitoring system among them was developed based on the embedded system in order to provide real-time information around a car in blind spots and narrow spaces. In this paper, we have implemented an around view monitoring system which enables image storing by combining the strong points of the around view monitoring system and those of the black box. In this system, location information which is not provided by the existing AVM systems of automotive companies is added to the around view monitoring system based on Car PC by way of using a USB GPS receiver with which location information and speed can be detected. This study has implemented an image storing system in which latitude, longitude and speed information are detected by using the RMC sentence summarizing the minimum data of GPS through the GPS receiver, and images from four cameras are input and synthesized and then, saved altogether with location and speed information in the storing device.

This paper proposes a multicasting channel allocation policy in the proxy server that a multicasting channel is not allocated for all objects but for some objects whose access pattern statisfies some conditions. For an object, one unicasting channel is allocated from the origin server to the proxy server and subsequently the proxy server either unicasts or multicasts that object from the proxy server to the client(s) based on the number of avtive streams. The proxy server also caches each object partially or fully based on each object’s access frequency : the cached portion of each object increases as its access frequency increases whereas the cached portion of each object decreases as its acces frequency decreases.

In this paper, for the implementation of the recognition system of car license plates, the region extraction algorithm based on the contour tracking and the new enhanced neural networks learning algorithm are proposed, which extracts the areas of car license plate and the character areas from the car images and recognizes the car license numbers from the extracted areas. And a candidate area was selected, whose density rate was corresponding to the properties of the car license plate obtained in the condition of the car license plate. The contour tracking algorithm extracted the feature areas covering the areas of characters from the car license plate. As well, the enhanced neural networks learning algorithm, combining the modified ART1 and supervised learning algorithm, recognized the car license numbers from the feature areas.

In this paper, we propose a dynamic multicasting policy based on the phased proxy caching to reduce the network bandwidth from the central server to the clients. In the proposed multicasting policy, for an object, a multicasting channel is allocated only if there are more than two active unicast streams for that object. So, the length of a multicasting window varies with the arrival rate of that object and moreover, proxy caching is applied to the multicasting policy by caching objects step by step starting from the initial portion of the objects; for each object, caching is done either partially or fully depending on the streams' arrival rate.