摘要
基于二元树复数小波变换的优良性能,结合遗传算法及证据理论,对图像纹理特征进行分析处理.首先利用DT-CWT的移动不变性、良好的方向选择性、完备的重构性、有限的与尺度无关的冗余度以及很高的计算效率等特点提取丰富的图像纹理特征;然后应用遗传算法的寻优能力对纹理特征进行约简;最后采用基于证据理论的数据融合方法实现多源信息融合目标识别.物体非损伤性评价实验表明该算法具有很好的识别效果.
Based on the excellent properties of dual-tree complex wavelet transform （DT-CWT）, the image texture features are analyzed and treated by combining the genetic algorithm and the evidence theory. Firstly, such characteristics of DT-CWT as shift invariance, excellent directional selectivity, perfect reconstruction, limited redundancy independent of the number of scales and high computation efficiency are utilized to extract abundant image texture features. Secondly, the global optimization ability of genetic algorithm is introduced to reduce the texture feature set. Thirdly, the data fusion method based on evidence theory is adopted to realize multi-source information fusion and object identification. The nondestructive evaluation experiment proves the high performance of the proposed algorithm.
出处
《信息与控制》
CSCD
北大核心
2007年第2期160-164,共5页
Information and Control
基金
教育部博士点基金资助项目(20060286005)
关键词
信息融合
目标识别
二元树复数小波变换
遗传算法
证据理论
information fusion
object identification
dual-tree complex wavelet transform
genetic algorithm
evidence theory
