We determined the performance of an optical correlator that used input
images of N x N pixels to examine larger images of LN x LN pixels. Th
e larger images were examined in parallel using a pyramid technique. W
e used morphological processing to approximate a low-pass filter to ge
nerate an image pyramid for a binary correlator. In correlation experi
ments we found that the discrimination between objects and SNR dropped
with increasing L. Although the discrimination may remain useful, the
more rapidly decreasing SNR limited the value of L that could be used
. In addition, we showed by simulation the processing of a 512 x 512 p
ixel image from an infrared sensor with a 128 x 128 pixel binary optic
al correlator.