CAJ | 학술논문

传统的基于视频的交通流量统计算法需要对流量统计区域属性及换算参数进行设置，设置方法复杂繁琐，灵活性差，应用有局限性。针对这一问题，结合人类统计目标数目的思想，提出了一种基于视觉理解的交通流量统计算法，无需设定流量统计区域就能完成流量统计。该算法使用帧间差分法检测出动态目标，通过 Camshift 与 Kalman 滤波相结合的方式完成目标跟踪，再通过对跟踪目标运动的分析来对交通流数目进行校正更新，最后获得交通流量数据。实验表明，提出的算法具有较好的统计效果。
전통적기우시빈적교통류량통계산법수요대류량통계구역속성급환산삼수진행설치，설치방법복잡번쇄，령활성차，응용유국한성。침대저일문제，결합인류통계목표수목적사상，제출료일충기우시각리해적교통류량통계산법，무수설정류량통계구역취능완성류량통계。해산법사용정간차분법검측출동태목표，통과 Camshift 여 Kalman 려파상결합적방식완성목표근종，재통과대근종목표운동적분석래대교통류수목진행교정경신，최후획득교통류량수거。실험표명，제출적산법구유교호적통계효과。
The traditional statistics algorithm for traffic flow based on video needs settings for the properties of the statistics area and the conversion parameters, which is complex, inferior in flexibility and also has a limitation. This paper proposes an algorithm of counting the traffic flow based on vision understanding, combining human′s thoughts of counting the number of targets, without setting the traffic flow counting area. In the algorithm, the frame difference method helps to detect the moving targets, and the combination of Camshift and Kalman filter can accomplish the tracking. Then we can correct the counting result and provide the traffic flow data as well through the analysis of the targets tracking. The result of the experiment shows that the algorithm functions well.