모바일 GPGPU에 최적화 된 데이터 전송기법을 적용한 병렬 차선 인식 알고리즘의 구현

In this paper, we implement Lane Detection algorithm for automobile with parallel processing on mobile GPGPU. This algorithm apply Gaussian Filter and Sobel Algorithm to a roadway image, and it apply Hough Transform to area of roadway that we expect there exist traffic lane. For a verification we use a Mobile GPGPU that has 100 MHz operation frequency and 4 core / 48 thread, and implemented Gaussian and Sobel algorithm run by parallel processing.