Human-Like Interpretation of Lines Using Embedded GPU

Abstract

In the applications where human like interpretation of shapes are required, the Hough Transform is popular because of its noise immunity and simple algorithm.Due to the computational capabilities and power consumption, the real time implementation of this algorithm on a standard CPU is not a good practice. In this paper, we have done line detection using Hough transform which is implemented on NVIDIA Jetson TK1 GPU in order to achieve less execution time with low power consumption. We have used the property of GPU to accelerate the process of line detection and compared the performance of CPU and GPU. We have shown different application of line detection which is implemented on Embedded GPU.