摘要
提出了一种融合纹理和颜色模型的改进mean-shift算法,可以克服传统目标跟踪方法中目标特征单一的缺点。在传统mean-shift算法的框架下,通过融合目标信息的颜色模型和Contourlet变换后的纹理模型,从多角度建立匹配特征,并在嵌入式平台上实现本文提出的算法。实验结果显示,跟踪效果得到明显的改善。
The improved mean-shift algorithm based on texture and color model is proposed, which can overcome the disadvantage of sin gle target feature in the traditional target tracking. In the framework of traditional mean-shift algorithm, the system establishes the matc hing feature from multiple perspectives, by fusing color model and texture model obtained from Contourlet transform. The proposed algo rithm has been realized on an embedded platform. The experimental results show that the tracking effect is improved.
出处
《单片机与嵌入式系统应用》
2015年第7期10-13,共4页
Microcontrollers & Embedded Systems
