Yazar "Uzer M.S." seçeneğine göre listele

Listeleniyor 1 - 3 / 3
  • Küçük Resim Yok
    Öğe
    A new hybrid feature selection method based on association rules and pca for detection of breast cancer
    (2013) Inan O.; Uzer M.S.; Yilmaz N.
    In this study, a new hybrid feature selection method named as AP has been formed to detect breast cancer, using association rules (Apriori algorithm) and Principal Component Analysis (PCA) together with artificial neural network classifier. Thanks to this hybrid system, both the decrease in the size of data and the successful and fast training of classifiers have been achieved. In order to detect the accuracy of the suggested system, Wisconsin breast cancer data have been used. 10-fold cross-validation has been used on the classification phase. The average classification accuracy of the developed AP + NN system is 98.29%. Among the studies performed through cross-validation method for breast cancer, our study result appears to be very promising. As the results suggest, this system, which is performed through size reduction, is a feasible system for faster and more accurate diagnosis of diseases. © 2013 ICIC International.
  • Küçük Resim Yok
    Öğe
    A real-ti?me tracking application of different coloured objects with a vision based mobile robot [Görme tabanli mobi?l robot i?le farkli renklerde nesneleri?n gerçek zamanli taki?bi?]
    (2010) Uzer M.S.; Yilmaz N.; Bayrak M.
    In this study, a real-time tracking application of different colored objects with a vision based mobile robot has been implemented. The developed robot has some features that can be used in the fields of exploration, security and observation. This vision based mobile robot is autonomously operated by using image processing and robot vision techniques. The response time of the robot that can track different color (red, blue and green) objects is between 96 ms and 106 ms as suitable for real time processing. By two different developed algorithms, for blue and red color objects a %100, and for green objects a %60 recognition success is seen with experiments. In colorful object tracking experiments in linear and circular orbitals, for an average value of 5.7 cm/s speed, a 4.5 cm deviation as maximum is confirmed.
  • Küçük Resim Yok
    Öğe
    A real-time object tracking by using fuzzy controller for vision-based mobile robot
    (2011) Uzer M.S.; Yilmaz N.
    In this study, a real-time object tracking application based on robot vision is presented. The developed robot has features that enable it to be utilized also in expedition, security, object-based indoor navigation and observational activities. Enabling the autonomous motion of this vision-based mobile robot, by just using robot vision methods and fuzzy logic control was the aim of the study. A vision-based object tracking algorithm, conventional PD control algorithm and fuzzy control are used and compared in study. The robot system focused on fuzzy control and image based control which is useful as a test bench tool for graduate and undergraduate students who are interested in robot control and image processing. © 2011 Academic Journals.