摘要
鉴于传统混合高斯模型在光照突变、噪声干扰时鲁棒性不高,易造成检测错误等问题,提出了一种改进的视频运动目标检测算法。该算法将混合高斯模型与六帧差分算法相结合,构建了一种高效的运动目标轮廓模型,并嵌入背景替换法和动态阈值分割法提高算法的稳健性,通过连通性检测和形态学处理,得到完整的运动前景像素。不同场景的视频检测结果表明,改进算法有效克服了光照突变、噪声干扰、空洞及双影现象,与同类算法相比,具有更高的准确度和鲁棒性。
As the traditional Gaussian mixture model has the shortcomings of lower robustness under the illumination change and noise interference,an improved video moving object detection algorithm is presented. Based on Gaussian mixture model and six-frame difference,an efficient moving object contour model is established. The stability of the algorithm is enhanced by embedding background replacement method and dynamic threshold segmentation method,and the complete moving foreground pixels can be gotten through connectivity tests and morphology processing. Video object detection results in different scenarios show that this algorithm can effectively solve the problems,such as the illumination abrupt change,noise interference,light cavity and double phenomenon. Compared with the other algorithms,the improved algorithm has higher robustness.
出处
《激光与红外》
CAS
CSCD
北大核心
2016年第2期240-244,共5页
Laser & Infrared
基金
国家自然科学基金项目(No.61263036)资助
关键词
运动目标检测
混合高斯模型
六帧差分
动态阈值
moving object detection
Gaussian mixture model
six-frame difference
dynamic threshold
