A FAST AND ENERGY EFFICIENT PARALLEL IMAGE FILTERING_x000D_ IMPLEMENTATION ON RASPBERRY PI'S GPU

Abstract

This paper presents a powerful processing technique for fast and energyefficient image filtering algorithm focusing energy and time-sensitive_x000D_ embedded and robotic platforms. Digital video processing is getting more and_x000D_ more popular in battery-powered devices like mobile robots and smartphones_x000D_ whereas in most cases, it leads overhead on the main central processing unit_x000D_ (CPU) and it consumes a significant amount of energy from the battery. It is_x000D_ suitable for parallelism since there is no data dependency between the steps_x000D_ of the two-dimensional convolution algorithm. We propose a vector version_x000D_ of the two-dimensional convolution algorithm, which can run parallel on_x000D_ embedded processors that has general purpose graphic processing unit_x000D_ (GPGPU), to reduce computation time and energy consumption. Our in-depth_x000D_ experiments shows that using GPGPU could reduce the execution time while_x000D_ guaranteeing lower power consumption and offloading the system CPU._x000D_ Experimental results showed that we achieved up to 105 times faster operation_x000D_ and 100 times less energy consumption compared to the CPU implementation._x000D_ Besides, we reduced the CPU overhead up to 10 times.