Synthetic Discriminant Filter Research Articles

Integrated zigzag Vander Lugt correlators incorporating an optimal trade-off synthetic discriminant filter for invariant pattern recognition

Optical correlators (OCs) have a wide range of applications in the field of pattern recognition. We present a type of integrated zigzag Vander Lugt correlator (IZVLC). The IZVLC incorporates two Fourier transform lenses and one spatial light modulator (SLM) as a programmable filter, which can effectively miniaturize the volume of OCs and hence achieve better optical pattern recognition. In order to precisely recognize distortion targets, an optimal trade-off synthetic discriminant filter (OTSDF) is programmed on the SLM of the IZVLC. Both the simulative and experimental results have shown that the IZVLC can effectively identify distortion targets within a rotation angle of ±π/3 and a scaling range of 0.6–1.6.

Nonlinear transformation analysis of the misregistered trade-off heterocorrelation filter: part 2, using the phase-only filter

We continue our study of the misregistered trade-off heterocorrelation filter (HCF), introduced in Part 1. Instead of using the matched filter, we use only the phases of the Fourier transforms of the functions constituting the filter, with our basic filter being the phase-only filter. We show that by adjusting the value of the trade-off factor it is possible to make the correlation pattern and intensities of the heterocorrelation peaks equal to those of the autocorrelation peaks; in effect, making the HCF a homogeneous filter. This means that it can equally recognize totally different objects that one designs it to recognize. These results (1) turn out to be independent of the amount of misregistration (i.e., shift) between the centers of the impulse responses of the functions making up the filter and (2) can also support the claim that a HCF is a totally new approach to generating synthetic discriminant filter functions. We also produce plots showing how the intensities, peak-to-noise ratios, and peak-to-secondary ratios of both autocorrelation and heterocorrelation peaks behave as a function of trade-off factor.

Family of heterocorrelation filters

We introduce a new concept in pattern recognition that we call heterocorrelation. Contrary to standard approaches, heterocorrelation allows correlation of different images solely by modifying the filter's intensity transmissivity in the areas of greatest phase mismatch relative to the phase of the stored template. This approach can convert a single object recognition filter to a multiple object classification filter. We develop three algorithms and successfully test each of them for three completely different input cases: one that uses geometric input images with very little common edge information, one that uses these same images encoded with high spatial frequency information, and one that uses synthetic aperture radar (SAR) images that have almost no edge information. The third algorithm provides shift-invariant heterocorrelation with equalized in-class correlation peaks, and eliminates the possibility of any higher-order side peaks for the added in-class objects. We also demonstrate how the use of a heterocorrelation filter can improve the performance of conventional multiplexed filters, such as synthetic discriminant filters, as well as increase their template storage.

Nonlinear filter for pattern recognition invariant to illumination and to out-of-plane rotations.

Automatic target recognition in uncontrolled conditions is a difficult task because many parametersare involved. This study deals with the recognition of targets under limited out-of-plane rotations while maintaining invariance to ambient light illumination. Contrast invariance is achieved by using the recently developed locally adaptive contrast-invariant filter, a method that yields correlation peaks whose values are invariant under any linear transformation of intensity. To reduce the sensitivity to the orientation of the object we replace the reference in the nonlinear filter by a synthetic discriminant filter. The range used for out-of-plane rotations was 40 degrees with a depression angle of 20 degrees. We present results for unsegmented targets on complex backgrounds with the presence of false targets.

Angle determination of airplanes by multicorrelation technique with optimal trade-off synthetic discriminant filters

The performance of optimal trade-off synthetic discriminant filters is illustrated in a particular application that consists of determining the angle and position of an airplane. A real-time numerical experimental system was realized using different airplane scale models. The performance of the system in response to different kinds of input perturbations is shown. The system allows us to obtain a 0.1-deg mean precision in the absence of noise. The mean precision is still better than 1 deg with an input SNR of about -22 dB. These results are obtained for 128- x 128-pixel input images. The system performance clearly demonstrates the effectiveness of the correlation technique for pattern recognition.

Properties of computer-generated, phase-only synthetic discriminant filters

The paper presents the experimental properties of the phase-only synthetic discriminant functions (SDF) filter when trained for invariance to in-plane rotation. Digital simulations and optical implementations by computer-generated holography are presented with results showing rejection of objects with a mean intensity six times that of the training class.

A complex synthetic discriminant filter for the parallel recognition of color patterns

We propose to combine the properties of color image correlation with the flexibility of synthetic discriminant filters to generate filters capable of solving various color pattern recognition problems. We show that in addition to being color sensitive, the resulting complex composite filters yield an improvement in recognition accuracy with respect to conventional intensity filters.