大视场双目主动视觉传感器的协同跟踪方法

针对大视场视觉监控的应用,提出了一种基于球面坐标的双目主动视觉传感器协同跟踪方法。首先采用图像特征匹配的方法估计摄像机的内部参数,然后引入球面经纬坐标系作为双目视觉传感器的公共坐标系,最后结合场景深度范围实现目标的协同跟踪。对于任意安装的双目视觉传感器,提出了一种建立球面公共坐标系的新方法,统一了不同参数下的情形,可以实现任意参数下的协同跟踪。实际监控场景下的视频实验,验证了方法的有效性和可行性。

Collaborative object tracking algorithm in large FOV with binocular active vision sensors

Active vision sensors have been widely used in surveillance domain bec ause of their ability to obtain multi-view -angle and multi-resolution information.Aiming at the application of visual s urveillance in large field of view (FOV),a novel collaborative object tracking algorithm for binocular active vision sensors base d on spherical coordinate model is proposed. Firstly,the feature matching based method is applied to calculate the intrinsic parameters of each camera.Then,the spherical longitude and colatitude coordinates are introduced to construct commo n coordinate system for the binocular sensors.Finally,the cooperative object tracking can be realized using the dept h range of surveillance scene.Furthermore,a novel procedure has also been developed to establish common spherical coordinate system for arbitrary installation of binocular sensors.The proposed method is independent of the parameters of activ e sensors,which can help to achieve cooperative tracking at arbitrary parameter values.Experiments are conducted in real surveillance environment and extensive results demonstrate the effectiveness and practicability of the proposed method.The proposed method can be used in visual surveillance with wide application prospect.