Pattern recognition with an adaptive joint transform correlator

An adaptive joint transform correlator for real-time pattern recognition is presented. A reference image for the correlator is generated with a new iterative algorithm. The training algorithm is based on synthetic discriminant functions. The obtained reference image contains the information needed to reliably discriminate a target against known false objects and a cluttered background. Calibration lookup tables of all optoelectronics elements used are included in the design of the adaptive joint transform correlator. Two methods for the implementation of the proposed joint transform correlator in an optodigital setup are considered. Experimental results are provided and compared with those of computer simulations.     

Image classification with a chirp-encoded joint transform correlator

We describe a method of performing image classification with a chirp-encoded joint transform correlator. In the proposed system the reference images and the input image that is to be classified are placed in different input planes of the joint transform correlator. As a result, different output planes of the correlator are associated with each reference image. The input image is classified on the basis of the intensity and the spatial position of the correlation peak. The reference images and the input image can be positioned in one input plane with glass blocks of different thicknesses placed on each reference image. This produces the same effect as having the reference images and the input image in different planes. Analytical expressions, computer simulations, and optical experiments are presented to investigate the performance of the chirp-encoded joint transform correlator for image classification.     

Modified fringe-adjusted joint transform correlation to accommodate noise in the input scene

A modified fringe-adjusted joint transform correlator is proposed that is able to accommodate noise in the input scene. The effect of noise in the input scene on the performance of the joint transform correlator is analyzed and quantified. When the target is embeddedin aseverely noise-corrupted input scene, it is shown that the proposed modified fringe-adjusted filter joint transform correlator delivers a better correlation performance and the capacity to accommodate this noise than does the fringe-adjusted filter-based correlator. When the power spectra of the input image and the reference image are subtracted from the power spectrum of the joint-input image, it is found that the noise effect on the output plane is independent of the objects in the input scene and originates from the convolution of the reference image and noise in the input scene.     

Optical wavelet transform by the phase-only joint-transform correlator

A method is presented that performs the optical wavelet transform with liquid-crystal televisions as spatial light modulators operating only on the phase of the incident coherent light. The architecture is the joint-transform correlator, and the wavelets and the image to be transformed are encoded in the input plane of the system. The mathematical formalism describing the adaptation of the joint-transform correlator to the wavelet transform is given and extended to the operation of the phase-only joint-transform correlator. A new wavelet is described for two-dimensional image processing, and experimental results are presented for optical wavelet transforms done in real time by use of this wavelet in the phase-only joint-transform-correlator architecture. The analysis is extended to multiwavelet (multispectral) analysis by the joint-transform correlator, and simulation results are given. Finally experimental results with the phase-only joint-transform correlator applied to multi-wavelet analysis are presented.     

抗噪性联合变换相关器

我们对联合变换相关器的输入图象进行Roberts梯度处理,同时对参考图象采取相移技术,并在频域中将功率谱二值化。与传统的联合变换相关器（CJTC）相比,新型的联合变换相关器（NJTC）消除了功率谱的直流分量,锐化了相关峰强度,提高了衍射效率和识别能力,同时增强了联合变换相关器的抗噪性能。用计算机模拟出预言结果。     

Adaptive binary joint transform correlator for image recognition

Our research has shown that the autocorrelation peaks of a binary joint transform correlator are affected by input scenes' backgrounds. An adaptive method is proposed to overcome this problem. The image of interest is first extracted from the background based on the position of the highest correlation peak of the input and reference images. The extracted image is then correlated with the reference to obtain the final correlation peak. Numerical simulations showed that the final autocorrelation peak is the maximum constant for a specified reference image.     

Multiple-target detection with use of a modified amplitude-modulated joint transform correlator

Multiple-target detection with a modified amplitude-modulated joint transform correlator is proposed. With this technique the joint power spectrum is first modified by the subtraction of the power spectrum of the input scene only and of the reference image from it; the resultant modified joint power spectrum is next multiplied by the amplitude-modulated filter function. The effect of noise in the input scene on the performance of the joint transform correlator is analyzed and quantified. This technique is found to deliver a correlation output and the capacity to accommodate noise in the input scene better than both the fringe-adjusted filter-based joint transform correlator and the modified fringe-adjusted filter-based joint transform correlator.     

Complementary-reference joint transform correlator

We present a new input configuration for the joint transform correlator called a complementary-reference joint transform correlator (CRJTC). The difference between the CRJTC and the existingjoint transform correlator is that the CRJTC has an additional complementary reference in the input plane. We use a criterion defined as the ratio of the cross correlation between the target and the reference to the cross correlation between the target and the complementary reference to perform recognition. This ratio can attain its maximum if and only if the reference and the target are matched and if it is stable to the light-source intensity fluctuations; therefore it is unnecessary to normalize the input images. The CRJTC is suitable for the recognition of a binary-amplitude image only. The experimental demonstration of a CRJTC is given.     

Optical implementation of neural networks for face recognition by the use of nonlinear joint transform correlators

We describe a nonlinear joint transform correlator-based two-layer neural network that uses a supervised learning algorithm for real-time face recognition. The system is trained with a sequence of facial images and is able to classify an input face image in real time. Computer simulations and optical experimental results are presented. The processor can be manufactured into a compact low-cost optoelectronic system. The use of the nonlinear joint transform correlator provides good noise robustness and good image discrimination.     

All-optical nonlinear joint Fourier transform correlator

We present the first all-optical nonlinear joint transform correlator based on a square-law receiver in the Fourier plane. Our device uses a photorefractive limiting quadratic processor. The compressional nonlinearity associated with the transfer function of the limiting quadratic processor enables the correlator to detect signals embedded in Gaussian and non-Gaussian noise. In the limiting region this device correlates the phase-only information of the input. This is the first time to our knowledge that photorefractives or real-time holography has been used in the correlation of the phase-only information. We demonstrate the operation of this device experimentally, and we evaluate its performance throughcomputer simulation for various forms of noise.     

Phase-shifting joint transform correlation with phase-iterative algorithm: effect of the dynamic range limit

With phase-shifting joint transform correlation (PSJTC) one uses multiple phase shifts to recover the phase difference between Fourier transforms of the input and the reference. In PSJTC systems the resulting phase-only function is used instead of the joint transform power spectrum (JTPS). Provided it can be recorded linearly, the JTPS reduces to a modulated sinusoidal fringe, especially when the target image matches the reference image. In practice, the JTPS has a wide dynamic range, and a CCD camera has a nonlinear response. Correspondingly, the recorded JTPS turns out to be different from a perfect sinusoidal fringe. Here we study the dynamic range effects of realizing PSJTC with the phase-iterative algorithm.     

Security and encryption optical systems based on a correlator with significant output images

An improved optical security system based on two phase-only computer-generated masks is proposed. The two transparencies are placed together in a 4f correlator so that a known output image is received. In addition to simple verification, our security system is capable of identifying the type of input mask according to the corresponding output image it generates. The two phase masks are designed with an iterative optimization algorithm with constraints in the input and the output domains. A simulation is presented with the resultant images formed by the two phase-only elements. Various mask combinations are compared to show that a combination is unique and cannot be duplicated. This uniqueness is an advantage in security systems.     

Performance of a phase-transformed input joint transform correlator

Conventionally a detected image is represented by an intensity array owing to the square-law nature of most detectors. However, this does not mean that we have to restrict ourselves to using intensity images for the correlation process. Transforming intensity images into phase images before correlation, which can be easily realized by a phase-modulation spatial light modulator, offers an alternative approach for high-performance pattern recognition. A phase-transformed input joint transform correlator is investigated in detail in terms of pattern discriminability, detection efficiency, and noise robustness. We show that the phase-transformed joint transform correlator has higher pattern discriminability and detection efficiency than the conventional joint transform correlator, and it also offers a better trade-off between the pattern discriminability and noise tolerance. A proof-of-concept experiment is also provided.     

Compact phase-conjugating correlator: simulation and experimental analysis

A simulation and experimental investigation of a recently proposed, compact, phase-conjugating correlator is undertaken. The effects of noise and other distortions in the input image and in the correlator filter plane are considered. As with other phase-only designs, the phase-conjugating correlator is sensitive to distortion of the input image while being robust in the presence of filter-plane distortions; this robustness is enhanced by the phase-conjugating property of the design.     

All-optical subtracted joint transform correlator with a holographic interferometer

A new, to our knowledge, subtracted joint transform correlator (SJTC) is proposed that has no digital processing in a computer. All processing for obtaining correlation signals between an object and multiple reference patterns were treated optically by use of a joint transform correlator with a holographic interferometer similar to the Mach-Zehnder one. The joint power spectrum of the reference patterns was subtracted from that of the input image (the object pattern plus the reference patterns), and the spurious correlation signals between the different reference patterns were removed. Because of the optical parallel computations of the Fourier spectra and the subtraction, the real-time SJTC is possible to achieve by use of only an optical system. An experimental arrangement of the system and system performances of the shift-invariant characteristics and discriminability are described. The results show the good performance of this system.     

Performance improvement of a phase-shifting joint transform correlator by use of phase-iterative techniques

We propose a new technique to improve the performance of the joint transform correlator (JTC). In this technique we have applied the phase-iterative algorithm to a phase-shifting JTC (PSJTC). By doing so, we restrain the noise that is contained in the recovered phase of the joint transform power spectra for the input images with background and additive noise. In the case in which the input image is embedded in the input noise, we find that, by using the phase-iterative techniques with the PSJTC, one can get a higher cross-correlation peak and signal-to-noise ratio than with a PSJTC alone. From the computer-simulation results, one can conclude that the proposed algorithm successfully enhances PSJTC performance, especially for an input image with large noise.     

Fresnel transform-based correlator

We present a new technique for information processing using Fresnel transform-based correlation. The main emphasis is on the design of a correlator involving the reference object and its near-field diffraction pattern at an optimized distance. The input-scene image and its diffraction pattern constitute the input pattern of the new correlator. The new technique shows a significant increase in discrimination ability and optical efficiency. Moreover, different encoding methods, such as the phase-only filter or the matched filter, can be used in conjunction with this method. A theoretical analysis as well as examples are given.     

Optical pattern recognition in motion acquired scenes using a binary joint transform correlation

Abstract

The problem of detecting patterns in moved scenes is analysed. The correlation when the input scene presents a relative motion with respect to the recording system image is studied analytically and an original method for detecting patterns by means of a binary joint transform correlator is presented. The method is based on the compensation of the sign errors introduced in the joint power spectrum by the transfer function of the degradation. Two alternatives to determine the sign of the transfer function are demonstrated: the first is based on an algorithm to extract information from the Fourier spectrum of the blurred image and the second method determines the sign errors by post-processing the correlation. Digital simulations and optical results are provided.     

Feature-extracted joint transform correlation

A new technique for real-time optical character recognition that uses a joint transform correlator is proposed. This technique employs feature-extracted patterns for the reference image to detect a wide range of characters in one step. The proposed technique significantly enhances the processing speed when compared with the presently available joint transform correlator architectures and shows feasibility for multichannel joint transform correlation.