利用多分辨分形特征进行目标识别
Target recognition based on multiresolution fractal feature
-
摘要: 该文分析了湖底回波信号经小波多分辨分解后,发现不同分解尺度上的小波系数的能量分别与分解尺度在一定范围内满足幂指数关系,与分形维的定义相似。据此求得两类特征参量,并对五类回波进行了分类,取得了较为理想的识别结果。
-
关键词:
- 多分辨分析; 分形; 识别
Abstract: In this paper, echoes are decomposed based on wavelet multiresolution analysis. It is shown that the energy of wavelet coefficients is exponential to the decomposed scale in some degree, which is similar to the definition of fractal dimension. Hence, two sorts of features are extracted and applied to recognizing five classes of echoes. The recognition results are satisfactory. -
陈捷,基于分形与混沌理论的水下目标特征提取研究[D],[博士论文],西安,西北工业大学,2000.[2]赵松年,熊小芸,子波变换与子波分析[M],北京,电子工业出版社,1997,36-80[3]王东生,曹磊,混沌、分形及其应用[M],合肥,中国科学技术大学出版社,1995,138-155.[4]吴高洪,章毓晋,林行刚,利用特征加权进行基于小波变换的纹理分类[J],模式识别与人工智能,1999,12(3),262-267.
计量
- 文章访问数: 2030
- HTML全文浏览量: 125
- PDF下载量: 560
- 被引次数: 0