离散信号尺度变化率及其在视频中的应用

Scale Rate of Change for Discrete Signals and its Application in Video Analysis

针对离散信号复杂度的定量描述问题,基于分形计盒维数的思想,将分形维数方法推广到适合离散信号数据的复杂度特征计算。提出尺度变化率来度量离散信号的局部复杂度,通过实验分析了具有不同时间变化率信号的局部复杂度特征,表明尺度变化率对信号局部特征描述的有效性。同时,将尺度变化率应用于数字视频序列处理,通过计算各像素灰度随时间变化的复杂度特征,实现有效的运动目标区域检测和分割。实验结果表明,基于尺度变化率的方法和已有方法相比,有相对更好的运动区域检测结果,并且尺度变化率的计算量小,相对于已有的复杂度估计方法,更适合离散语音、图像、视频等数字信号的实时分析处理。

In this paper,the quantitative measurement of complexity for discrete signal is studied.The idea of fractal dimension estimation is extended to a novel method suitable for discrete signal data analysis.The scale rate of change is proposed to measure the degree of local complexity for digital signals.The relationship between the scale rate of change and the time rate of change is investigated experimentally,which indicates the effectiveness of the proposed method in complexity measurement.The proposed complexity feature is applied in motion detection and segmentation in video sequence.The scale rate of change is estimated for each pixel in a frame,based on which the moving objects can be effectively detected and segmented from the video frame.Compared with currently used methods,the proposed method obtains better result of motion segmentation.Moreover,the computation of the proposed method is relatively small compared with existing methods of complexity measurement,which is more suitable for the real-time analysis and processing of discrete speech,image and video signals.