摘要
在应用光电混合联合变换相关器实现运动目标相关识别的过程中,针对目标与模板由于运动所产生的差异和复杂背景影响相关器识别率的问题,采用了一种以高斯函数的一阶导数做小波多尺度边缘提取和形态学膨胀进行边缘加粗处理相结合的方法。这种方法可充分利用小波多尺度的特点,使得提取后的边缘在抑制噪声的同时,能保留更多的细节信息,明显改善复杂背景下运动目标的相关识别问题。对以低对比度和小目标为特点的动态序列进行光学相关实验,结果表明,这种方法能有效增强探测目标的相关峰亮度,验证了算法对复杂背景下运动目标识别的可行性。
In the application of achieving the correlation recognition of moving target with hybrid optoelectronic joint transform correlator (HOJTC), the differences resulting from the movement between the target and the template and the cluttered background influence the correlation ratio of the correlator. A combined method of wavelet multi-scale edge extraction which uses the first order derivative of Gaussian function and edge processing with morphological dilation is applied. This method can fully use the wavelet multi-scale characteristics. After extracting the edge, the noise can be suppressed and more details can be retained. It can improve the detection efficiency of the moving target in cluttered background apparently. The optical correlation experiments on a low contrast and small moving target show that this method can enhance the brightness of the correlation peaks and verify that the algorithm is feasible to the moving target recognition in cluttered background.
出处
《光学学报》
EI
CAS
CSCD
北大核心
2012年第1期62-68,共7页
Acta Optica Sinica
基金
总装备部预研局十一五基金资助课题
关键词
傅里叶光学
光学相关识别
小波多尺度边缘提取
运动目标
膨胀
复杂背景
Fourier optics
optical correlation recognition
wavelet multi-scale edge extraction
moving target
dilation
cluttered background