【作者】 刘卓夫； 廖振鹏； 桑恩方；

【Author】 Liu Zhuofu~(1,2) Liao Zhengpeng~(1,2) Sang Enfang~1 (1 College of Underwater Acoustic Engineering,Harbin Engineering University,Harbin,150001,China) (2 Institute of Engineering Mechanics,China Earthquake Administration,Harbin,150001,China)

【机构】 中国地震局工程力学研究所； 哈尔滨工程大学水声工程学院；

【摘要】 根据分形维数可以表示曲线的复杂程度的特性,并结合声纳图像自身的特点,提出了通过计算图像频谱曲线的分形维数进行图像识别的方法。该方法具有计算简单、物理意义明确、抗干扰能力强的优点,实验结果表明,该方法分类精度高且计算速度快,克服了传统的纹理识别方法识别率较低的缺点。更多还原

【Abstract】 Fractal dimension can represent the complexity of curves.According to the distinguishing feature of sonar images, this paper puts forward the method of sonar image recognition by calculating the fractal dimensions of frequent spectrum curves. The method is easy to carry out and has the characteristics of definite physical meaning and strong noise resistance ability.The experimental results show that the method has high classification accuracy and is fast in calculation.It solves the problem of low recognition rate by using traditional texture recognition methods.更多还原