基于小波变换的分层块匹配多目标跟踪方法

Arithmetic research based on wavelet transform multi-layer block matching multi-target tracking

机载的航空相机或弹载的红外CCD(ChargeCoupledDevice)摄像头都是安装在运动载体上的CCD相机,其连续拍摄的图像背景是运动的,很多情况下要从运动背景中检测其中的运动目标,并进行跟踪。为解决全搜索算法运算量大、搜索时间长的问题,根据小波域中各图像之间的相关性,提出一种基于小波变换的分层块匹配跟踪算法,能从运动背景中检测出多个运动目标,并可计算出多个动目标的形心坐标,绘出各动目标的运动轨迹。该算法与基本块匹配跟踪算法相比,克服了原算法运算量大,信噪比较低时匹配不准确等缺点。

Aero-camera on plane and the CCD(Charge Coupled Device) camera on missile are all fitted on moving carrier. They continuously shoot image, which background is moving. Sometimes we need to detect moving targets in the moving background, and track them. An arithmetic based on wavelet transform multi-layer block matching tracking is suggested. It can detect moving targets from the background, and can work out the figure center of the targets, and trace out the movement contrail of the targets. Compared with traditional block matching tracking arithmetic, it overcomes some disadvantages, such as a mass of operation, inaccurate matching in the low SNR(Signal Noise Ratio) image.