Browsing by Author "Amarasiri, H.A.N.B."

Now showing 1 - 3 of 3
  • No Thumbnail Available
    Item
    Eye Blink Detection by Image Processing to Prevent Vehicle Accidents
    (Uva Wellassa University of Sri Lanka, 2018) Nimanthils, K.S.H.; Amarasiri, H.A.N.B.
    Deaths and injuries due to road traffic accidents are extremely high in Sri Lanka. Detecting the drowsiness of the driver is one of the surest ways of measuring driver's fatigue. Driver's drowsiness is one of the major reasons which leads to these mishaps. This research describes 'Real Time Drowsiness Detection System', which could determine the level of drowsiness of the driver. This system considers both the closing of eyes and opening of the eyes. The eye blink of the driver is detected. If the driver's eye remains closed more than a certain period of time, the driver is said to be drowsy and alarm is sounded. Viola Jones Algorithm and Hough transformation algorithm are used for iris detection. Primary attention is given to faster detection and processing of data. Keywords: Drowsiness, Hough transformation algorithm, Viola Jones algorithm, Eye blink
  • No Thumbnail Available
    Item
    Object Sorting System Using Affordable Robotic Manipulator
    (Uva Wellassa University of Sri Lanka, 2018) De Silva, A.M.C.; Amarasinghe, C.; Amarasiri, H.A.N.B.
    In modern world, automation and robotic are at a high demand as both directly affect the rapid development in various industries. Also, quality and flexibility of products are becoming significant criteria. Robots will be the best solution for substitution of cost of labor wages and higher customer demands. Automation is the most preferred technique to replace the human labor that perform complex tasks in complex and risky environments. Pick and place operations are required in many of manufacturing processes, and it can be automated. Therefore, this study aims to sort and place different objects according to their shape and color using image processing techniques. The main objective is to provide a solution for any manufacturing process which require sorting based on shape, color or their combination. In this study, low cost, servo motor controlled, CNC machined 5 DOF robotic manipulator with electromagnetic end effector is used for the object sorting. The shape detection and color detection algorithms were implemented using MATLAB 2016 and a highquality USB web camera. The robotic manipulator is controlled through an Arduino Mega via serial connection to MATLAB. Since, higher processing power is required for the algorithms, a computer was used. A graphical user interface was developed with customization options. Objects were successfully detected based on shape and color and sorted using the robotic manipulator. Keywords: MATLAB, Arduino, Color, Shape, Sorting
  • No Thumbnail Available
    Item
    Obstacle Avoidance System For A Quadrotor UAV Using Over Head Mounted Camera, Based On Image Processing Technique
    (Uva Wellassa University of Sri Lanka, 2018) Prabhashitha, R.D.; Amarasiri, H.A.N.B.
    An obstacle avoidance system for a Quadrotor UAV using overhead mounted camera is based on image processing technique. Current robots use various sensors to obtain the locations of obstacles and navigate without colliding. Here, the robot uses an overhead camera to detect obstacles for safe navigation. For this method, the robot does not need to use any additional sensors to detect obstacles. Here, the method uses an image processing technique. The images from overhead web camera are converted to grayscale images and filtered to remove noise and then converted to black and white binary images by using MATLAB software package. Then, objects in the images are identified and centroids are drawn, and a square is drawn around the robot to distinguish in-between. Thereafter, the pixel location of each object is taken from the centroid pixel location and the area of the object is also taken into the algorithm. The navigation is done with the aid of the "depth first search" algorithm. It uses the location of the robot to navigate to the desired node in the image binary matrix. The results of this project were successful. Navigation of a mobile robot using overhead mounted single camera without using additional sensors were done successfully. For Quadrotor, this algorithm was simulated on Robotic Operating System and Dronekit Simulation platform. These platforms were built on Ubuntu operating System and a Python script was used to connect the MATLAB program. Simulation successfully detected the obstacles and had a safe navigation. The conclusion of the research is that, this method is suitable for applying for any mobile robot for a specific range and the method can be used for Quadrotor by defining same parameters as for the mobile robot. Keywords: Obstacle Detection, Quadrotor, Image Processing, MATLAB, Dronekit