基于多色彩标识的跟踪及交互方法

Method for Tracking and Interaction Based on Multiple Color Markers

为了增强标识的环境融入性,更加贴近自然特征,提出一种基于多色彩标识的跟踪注册和交互方法。对视频图像中不同色彩标识,采用一种运用自适应阈值来检测标识的方法进行轮廓检测,对轮廓图像进行校正,与模板图像进行图案匹配;为了能准确地获得目标,需要进一步对目标图像进行颜色匹配,提出一种新的基于L1L2L3和rgb颜色模型相结合的颜色匹配方法实现颜色匹配。根据匹配结果计算出摄像机的参数,完成虚拟物体的注册。应用颜色模型与LK方法跟踪角点相结合来解决物体的遮挡,实现了触摸式交互操作。通过实验证明,该方法基于多色彩标识能够准确稳定注册虚拟物体,交互操作便捷及准确,与ARTool Kit比较在对光照和环境的适应性上要有明显提高。

A method was proposed for real-time tracking and interaction based on multiple color markers, which provided enhanced integration of physical, virtual and augmented environments. For marking with different colors in video images, a marker detection method based on adaptive threshold was initially used to detect the marker＇s contour in the captured image. The detected contour was then corrected and the marker pattern is matched with the template image set. To accurately obtain the image target, further color matching was undertaken for the target image. A color matching method which combined the LIL：Ls and rgb models was proposed. The camera parameters were calculated by using data obtained from color and marker matching, and a virtual object was registered onto the marker in the physical environment. Furthermore, object occlusion was addressed by means of the color model and tracking corners via the LK technique in order to realize touch interaction. Experimental results show that the proposed method accurately registers virtual objects on markers in video images and is able to track objects in real-time. Interactive operation is both convenient and accurate by using the method. The method also adapts well to varying illumination and environments in comparison to the ARToolKit software.