Yunus Santur

IMU based adaptive blur removal approach using image processing for railway inspection

Rail transportation systems, which are commonly used in todays world, should be inspected at certain intervals for possible accidents. During the rail inspection, the physical vibration on rail lines causes a blurring effect on the images. Doing deblurring automatically requires information of blurring rates and specifying the parameters accordingly for deblurring. With this purpose, a test equipment which can move on rail lines and a camera system for the detection of blurring and deblurring is integrated with Inertial Measurement Unit (IMU) is promoted in this study. Then, with Attitude and Heading Reference System (AHRS) algorithm, the effect of blurring at the moment of the vibration is examined, point spread function (PSF) value is chosen dynamically and deblurring is achieved. In order to increase the accuracy rates of detection algorithms, a pre-treatment method is proposed for detecting the blur effect and removing it.

Chouqet fuzzy integral based condition monitoring and analysis approach using simulation framework for rail faults

Today, railway systems are one of the most preferred means of transportation worldwide. Some faults may occur on railway tracks due to several reasons and such faults may cause accidents. For this reason, railway tracks should be periodically inspected. In this study, a study was performed which was aimed to locate possible faults such as cracks, holes and wear on rail surfaces fast and with high accuracy levels in accordance with real systems. The study is made in two stages and distance values read on laser cameras constitute the input data of the system. Rail profiles marked as working and out-of-order are presented as input data of the system and a random forest is created for learning after a pretreatment on the data, four different feature extraction methods are obtained with attribute. In test phase, a second attribute is obtained via feature extraction methods, classified and produced after diagnosis. General accuracy rates of the system are increased with fuzzy integral method and a system which can work in real time and with high accuracy rates on railways with a physical tester was proposed.

Smart pregnancy tracker system using social knowledge networks for women

Today, people use the internet extensively to meet their information needs, to socialize, to communicate, to handle formal and informal processes. The web-based information networks established for this purpose are growing day by day and reach a larger audience. Facebook, the worlds largest social network, has reached 2 billion users. There are social information networks established for different purposes as well. In this study, it was aimed to construct a social information network specific to women. It is to create a social information network specific to women targeted for work. In web-based network working with membership logic, members can access informative contents such as follow-up of processes such as period, pregnancy, baby vaccination schedule, body mass index, calculators such as baby percentile, tests, articles and visual aided trainings.

A new rail inspection method based on deep learning using laser cameras

Rail systems are one of the most important transportation methods used in todays world. The abnormalities that occur on railway tracks due to various causes result in breakdowns and accidents. For this reason, railway tracks must be periodically inspected. This study proposes a new approach for rail inspection. Today, the railway inspection process is generally performed using computer vision. But the oil and dust residues occurring on railway surfaces can be detected as an false-positive by the image processing software can lead to loss of time and additional costs in the railway maintenance process. In this study, a hardware and software architecture are presented to perform railway surface inspection using 3D laser camera and deep learning. The use of 3D laser cameras in railway inspection process provides high accuracy rates in real time. The reading rate of laser cameras to read up to 25.000 profiles per second is another important advantage provided in real time railway inspection. Consequently, a computer vision-based approach in which 3D laser cameras that could allow for contact-free and fast detection of the railway surface and lateral defects such as fracture, scouring and wear with high accuracy are used in the railway inspection process was proposed in the study.

Big data framework for rail inspection

It is necessary that periodical inspection and maintenance of railway transportation systems, which is becoming more and more common day by day, is required. A typical rail line can range from a few kilometers to thousands of kilometers. It is not possible to control this length of railway line with human labor. For this purpose, rail inspection is done automatically by machine vision today. The input data of machine vision systems consists of data from high-resolution cameras and other sensors. These data are evaluated by machine learning methods and the diagnosis result is produced. However, the data rate and the amount of data that occur in both long-distance and long-time repetitive ray inspection applications are huge. Proper handling, storage and analysis of this data requires a Big Data-based approach. In this study, an approach is proposed for the evaluation of large data obtained from vision-based diagnostic systems and the extraction of useful information in tracked systems. The proposed approach has been verified using simulation and experimental data and the effectiveness of the approach, utility, usability, and other visual-based diagnostic approaches to be developed in directed systems have been demonstrated.

Improving of personal educational content using big data approach for mooc in higher education

Today, web-based education technologies such as e-learning, distance learning, online course, virtual classrooms and interactive learning are commonly used outside the traditional education systems. Massive Open Online Courses, which were adopted with the evolution of these systems in 2008, have become the most popular education systems of today and access to a very large audiences with many modules such as video courses, documents, interactive learning activities and quizzes in themselves. In this study, a machine learning and big data based approach has been presented for the mentioned online education systems. With the proposed approach, it is aimed to develop course contents in online education, offer student-specific learning activities, perform an analysis according to criteria such as age, gender, occupation, education level and location, and to obtain decisions with strategic importance such as determining the course prerequisites to be developed by big data-based analysis.