چكيده به لاتين
Image processing is an ever expanding and dynamic area with applications reaching out into our everyday life such as medicine, space exploration, surveillance, authentication, automated industry inspection and many more areas. In the world of embedded systems, those applications and computer programs that are written for computer vision, should have enough performance such as PC in order that they could be implemented on embedded systems.
In the present study, an IR target detection algorithm will be implemented on PC and Raspberry Pi 2 board. This algorithm is a visual attention-based to detect the dim IR targets. The method selects difference of Gaussians (DoG) filters compute the saliency map and to reduce background noise and clutter. Target detection algorithm uses OpenCV library has been developed by the Intel Corporation. Then algorithm code is implemented on Raspberry Pi 2 using OpenCV and Qt. In the end, results of implementation of this algorithm in MATLAB and C++ qt-based on PC and C++ language Qt-based used on Raspberry Pi 2 is compared. The result show that the time of implementation (in C++) on PC almost 50 times its implementation with MATLAB and implementation of Raspberry Pi 2 increase the speed of the algorithm, which is almost 6.5 times. Multithread programing is used for optimization on the Raspberry Pi 2, which includes quad-core ARM Cortex-A7 processor to reduce execution time, as well as overclocked property on Raspberry Pi 2 is used for hardware acceleration. Experimental result show that the speed of algorithm is improved more than 40% after optimization.
