摘要
在离散分形布朗随机场（ＤＦＢＲ）理论的基础上，提出一种多分辨率目标识别算法。该算法克服了一般分形方法在固定尺度上提取图象分形特性的缺点，它以小波分解为主要数学工具，利用随机场的功率谱特性以及相邻两级小波分量之间的能量比关系，完成了由粗到精的目标识别。该算法的最大特点是对目标大小具有自适应性，特别适合于自然场景中的多目标识别，同时也使得计算量大大减少。
Based on the Discrete Fractal Brownian Random Field (DFBR) an algorithm of multiresolution object recognition is presented. This algorithm overcomes the defects of the general methods which extract the fractal characteristics within a fixed scale. Using wavelet as the major mathematical tool and by means of the power spectrum of DFBR and the rate of the wavelet coefficients between adjacent scales, a coarse to fine multiobject recognition is accomplished. This algorithm has the adaptability for the sizes of objects and requires less calculation.
出处
《数据采集与处理》
CSCD
1996年第4期246-248,共3页
Journal of Data Acquisition and Processing
基金
国家自然科学基金
关键词
图象识别
图象处理
小波分析
算法
image processing
object recognition
fractal
multiresolution
wavelet analysis
