基于金字塔分解的分层光学相关匹配策略

Layered Optical Correlation Matching Stragegy Based on Pyramid Decomposition

在光电混合相关识别系统中,由于光电和电光转换的限制,相关运算不能以光速来进行,降低了系统实际的处理速度。该文将金字塔分解用于光学相关匹配,以实现分层光学相关,来提高海量图像的处理速度。首先分析了构建图像金字塔过程中的滤波方法和目标尺寸对相关结果的影响,从而选定了合适的图像金字塔类型,确定了系统所容许的目标尺寸,然后阐述了分层相关匹配的实现方法。利用该方法处理速度至少可提高约4倍。

In hybrid photoelectric correlation recognition system, correlation operation can＇t be implemented at the speed of light velocity because of the restriction of photoelectric conversion and electro-optic conversion, which reduces the actual processing speed of system. Pyramid decomposition is applied to optical correlation matching to realize layered optical correlation, which can enhance the processing speed for massive images. At first, the effect on correlation results of tilting methods used in constructing image pyramid and the dimension of tar get, and the appropriate type of image pyramid and the dimension of target allowed by the system are determined. Then the realization method of layered correlation matching is discussed. The processing speed may be enhanced by about four times at least by using this method.