红外技术, 2019, 41 (3): 256, 网络出版: 2019-04-05
混合高斯融合三帧差的运动目标检测改进算法
An Improved Algorithm for Moving Target Detection Using a Gaussian Mixture with Three-frame Difference
混合高斯模型 三帧差法 目标检算法 测边缘检测 颜色空间 Gaussian mixture model three frame difference method target detection algorithm edge detection color space
摘要
针对混合高斯模型(Gaussian Mixture Model, GMM)无法检测到完整的运动目标, 三帧差法检测目标时对物体速度的敏感, 检测到的物体会出现空洞等缺点, 提出了一种混合高斯融合三帧差法的运动目标检测改进算法。首先, 在运动目标提取过程中, 改进的三帧差法采用动态分割阈值和边缘检测技术, 解决光线突变和边缘不连续问题; 然后引入新的高斯分布自适应选择策略, 以减少处理时间, 提高检测准确性; 最后, 利用改进 HSV(Hue-Saturation-Value)颜色空间来消除阴影区域, 得到一个完整的运动目标。数据实验表明, 该算法在不同场景具有较好的检测能力。
Abstract
For the mixed Gaussian model unable to detect the complete moving target, the three-frame difference method is sensitive to the speed of the object as the target is detected, and defects on the detected object (such as voids) appear. An improved moving target detection algorithm based on a mixed Gaussian fusion three-frame difference method is proposed. First, in the process of moving target extraction, the improved three-detect method uses a dynamic segmentation threshold and an edge detection technology to solve the problem of light mutation and edge discontinuity. Then, a new Gaussian distribution adaptive selection strategy is introduced to reduce processing time and improve detection accuracy. Finally, the improved HSV color space is used to eliminate the shadow area and obtain a complete moving target. Data experiments show that the algorithm has better detection capabilities in various scenarios.
于晓明, 李思颖, 史胜楠. 混合高斯融合三帧差的运动目标检测改进算法[J]. 红外技术, 2019, 41(3): 256. YU Xiaoming, LI Siying, SHI Shengnan. An Improved Algorithm for Moving Target Detection Using a Gaussian Mixture with Three-frame Difference[J]. Infrared Technology, 2019, 41(3): 256.