Enhancement of correlator filter performance utilising space-domain non-linearities.
It is necessary for a correla tor to be invariant to all distortions of a
reference object while still being able to maintain good discrimination between
similar objects. A filter must thus strike a compromise between the conflicting
requirements of in-class distortion tolerance and out-of-class discrimination. A
bandpass filter is therefore ideal; the high frequency response being attenuated
to provide less sensitivity to in-class variations while the low frequencies are
removed since they compromise the discriminatory ability of the filter.
This study proposes a nonlinear bandpass pre-processing operation applied to
both the input and the reference images prior to the correlation operation, i.e.
in the space domain rather than to the spectrum. This operation is shown to
result in greater tolerance to in-class variation while maintaining an excellent
discrimination ability. Additionally, the introduction of non-linearity is shown to provide greater robustness in the filter response to noise and background
clutter in the input scene.
Utilizing the good performance of this new filter and in order to provide
improved inter-class tolerance for a higher distortion range, the proposed
nonlinear bandpass operation is incorporated in the synthesis of a modified
Synthetic Discriminant Function (SDF) filter by applying the non-linearity to
each of the bandpass filtered training set images comprising the SDF. This
SDP filter is demonstrated to produce a sharp peak, and show excellent
discrimination ability, while maintaining invariance to out-of-plane rotation
over a distortion range of up to 90 degrees.
The processing sequence of this filter is suitable for implementation by a
hybrid digital/optical arrangement in which the input image is wavelet filtered
in real-time by a digital signal processor (DSP). The filter performance is
investigated with both input image and filter plane quantization to the limit of
binary amplitude and phase, and hence it is suitable for implementation with
current frame-rate Spatial Light Modulators (SLM) technology.