CRIIS LabMeeting 06/04/2018

Pranjali Shinde

The aim of the project was to classify the movement of the object and to detect imminent collision in real time. The objective of the thesis were to capture the video of the moving object and then process the frames one by one by forwarding it to the hard processor. The FPGA has an ARM Cortex A9 processor with computer vision running on the embedded Linux which aids in classifying the moving of the object. The frames captured by the camera at a rate of 30fps is in Blue, Green, Red (BGR) colour space which is converted to Hue Saturation Value (HSV) colour space which serves as an input for computer vision and image processing. Specific threshold values are set based on the colour of the object to be detected. Binarization is performed on the HSV frame where the presence of the object is denoted by ‘1’ and absence by ‘0’. The more number of 1s in the frame would determine the proceeding of the object towards the camera on which the looming detection would be performed. The looming detection is used to find imminent collision. Furthermore, if less number of 1s in the image it would determine the receding of the object from the camera. Classification of the movement of the object is performed based on the change in the x and y coordinate and contouring the object.