Distortion-invariant Pattern Recognition Research Articles

Distortion-invariant pattern recognition with local correlations

Local adaptive correlations for distortion-invariant and robust pattern recognition are presented. An iterative training algorithm for design of adaptive filters is utilized. The recognition performance of the proposed filters in noisy environment is compared with that of linear composite filters in terms of noise robustness and discrimination capability. Computer simulation results are provided and discussed.

Distortion-invariant pattern recognition using synthetic discriminant function based multiple phase-shifted-reference fringe-adjusted joint transform correlation

This paper proposes a novel pattern recognition system for invariance to noise and distortions. The technique first generates a synthetic discriminant function of the target image from its different distorted versions. It then takes four different phase-shifted versions of the reference image, which are individually joint transform correlated with the given input scene. Thus the proposed algorithm produces a single cross-correlation signal corresponding to each potential target. Also a fringe-adjusted filter is designed to generate a delta-like correlation peak with high discrimination between the signal and the noise. The pattern recognition system is also designed for the identification of multiple targets belonging to multiple reference objects simultaneously in a given input scene. The proposed technique is investigated using computer simulation including real-life images in different complex environments.

Bridging the fuzzy, neural and evolutionary paradigms for automatic target recognition

Multilayer perceptron neural networks possess pattern recognition properties that make them well suited for use in automatic target recognition systems. Their application is hindered, however, by the lack of a training algorithm, which reliably finds a nearly global optimal set of weights in a relatively short time. The approach presented here is based on implementation of genetic algorithms and fuzzy logic in training the proposed hybrid architecture. Compared to other approaches, it offers the following three main advantages. The neuro-computing technique is capable of fast and adaptive distortion-invariant pattern recognition for rapidly changing targets. On the other hand, genetic algorithms and fuzzy logic offer very sophisticated configuration control, which combines the results of previous computations with the external operating environment. Third, it allows us to significantly improve the reliability of object detection in the input scene with respect to associated distortions at no additional computational cost. This paper examines using genetic algorithms as an efficient way to train a feedforward neural net, the inputs for which are provided by a fuzzy front end, to be applied to automatic target recognition. The system is tested using actual laser detection and range data as training data and the results of the analysis show that the proposed system results in a much faster convergence on a weight set, and a high rate of successful recognition.

基于振幅调制纯相位滤波器和片状正交非线性广义相关的畸变不变识别

基于振幅调制纯相位滤波器和片状正交非线性广义相关的畸变不变识别

Distortion-invariant pattern recognition using synthetic discriminant function-based shifted phase-encoded joint transform correlator

A novel pattern recognition technique is proposed that employs shifted phase-encoded fringe-adjusted joint transform correlation (SPFJTC) for efficient real-time application and synthetic discriminant function for invariance to distortions. The proposed technique produces a single delta-like correlation for a potential target in the input scene, and performs successfully even in a noisy environment. The simple architecture of the proposed technique can be implemented easily on optoelectronics for very high-speed operation. Computer simulation results verify the efficient performance of the technique under different variations of the input scene and the environment.

Distortion-invariant multiple target detection using class-associative joint transform correlation

A distortion-invariant class-associative pattern recognition technique is proposed, where a class of objects may be defined as a group of objects with similarity and dissimilarity among them. The fractional power fringe-adjusted joint transform correlation technique as well as the synthetic discriminant function concept has been effectively utilized to achieve the distortion-invariant detection of multiple dissimilar targets simultaneously present in the input scene. Simulation results prove that the proposed scheme is an effective tool for the detection of multiple dissimilar targets in both binary and gray-level input scenes corrupted by distortion and noise.

Distortion-invariant pattern recognition based on a synthetic hit-miss transform

A new synthetic hit-miss transform (SHMT) algorithm is proposed for distortion-invariant recognition of various objects in noisy and cluttered input images. The proposed SHMT algorithm uses synthetic structuring elements (SEs), which are synthesized based on a synthetic discriminant function (SDF) filter algorithm. The synthetic hit SE is composed of the linear combination of the reference hit SEs, and the synthetic miss SE is composed of the linear combination of the reference miss SEs. Based on various simulations, it is shown that the proposed algorithm can be applied to an HMT correlator to improve its ability to detect various objects with distortions in noisy and cluttered scenes.

On a higher-order neural network for distortion invariant pattern recognition

This paper proposed some methods upon a second-order neural network. These networks apply the normalized frequency distribution of distance between two points on an object. 0.766 of the recognition accuracy for 2D two-class mixture distributions and 0.960 for hand-written characters are achieved.

Composite filters for inverse synthetic aperture radar classification of small ships

Pattern recognition of small ship inverse synthetic aperture radar (ISAR) images is considered. This represents a formidable new distortion-invariant pattern recognition problem. New weighted correlation range alignment and weighted multiple-scatterer motion compensation variations to standard image formation steps were developed and used. A new algorithm to determine if an input image is useful was developed and found to be necessary for small ship ISAR data. Initial recognition results using distortion-invariant (composite) filters are presented. These provide new concepts including: use of a validation set and a goodness measure to select filter parameters, use of new output criteria for which a filter in a bank of filters is best for class determination, rejection of decisions on some poor test input images, and use of voting over a time sequence of test inputs.

Joint transform correlator with minimum peak variance in nonzero-mean noise environment

The design of a nonzero order joint transform correlator with minimum peak variance has been presented for distortion invariant pattern recognition. We minimized the peak variance of the cross-correlation output in the presence of input nonzero-mean noise, subject to the constraints on the cross-correlation peak as well as the deterministic mean of the noise. A general solution for the space-domain based reference function has been proposed to remove the effect of the deterministic mean caused by the noise.

Minimum-variance nonzero order joint transform correlators

The design of the minimum-variance nonzero order joint transform correlator (MV-NOJTC) for distortion invariant pattern recognition is presented. We constrain the expected value of the cross-correlation peak to a constant for each input image in a training set, while minimizing the peak variance of the cross-correlation output due to input noise, a general solution for the space-domain based reference function of the MV-NOJTC is proposed.

Equal correlation peak optimization of the filter-feature-based synthetic discriminant reference image

A modified equal correlation peak (ECP) optimization is proposed and introduced into the filter-feature-based joint transform correlator (FFB JTC) used for distortion-invariant pattern recognition. Instead of one fixed and arbitrarily chosen point in the classical ECP optimization, the maximum point in the correlation output plane is selected in our method. The modified ECP optimization is quite efficient when involving filter modulation. When compared with other ECP optimization methods related to the filter-modulation, our optimization method is linear and can easily obtain an exact solution. Simulation results show that after ECP optimization the stability of the correlation output of the FFB JTC may be greatly improved.

Composite Fourier-plane nonlinear filter for distortion-invariant pattern recognition

We investigate composite Fourier-plane nonlinear filters for distortion-invariant pattern recognition. Different types of noise such as spatially nonoverlapping color background noise, other nontarget objects, real scene noise, and additive white noise are used to investigate the performance of the filter. Computer simulation results are presented to show the performance enhancement using Fourier-plane nonlinearities in the composite filter designs. The training image sets and the test image sets are generated by in-plane and out-plane rotation of different aircraft.

Composite versus multichannel binary phase-only filtering

Multichannel filtering and its inherent capacity for the implementation of data-fusion algorithms for high-level image processing, as well as composite filtering and its capacity for distortion-invariant pattern-recognition tasks, are discussed and compared. Both approaches are assessed by use of binary phase-only filters to simplify implementation issues. We discuss similarities and differences of these two solutions and demonstrate that they can be merged efficiently, giving rise to a new category of filters that we call composite-multichannel filters. We illustrate this comparison and the new filter design for the case of rotation-invariant fingerprint recognition. In particular, we show that the gain in terms of encoding capacity in the case of the composite-multichannel approach can be used efficiently to introduce multichannel-filter reconfigurability.

Projection-slice synthetic discriminant functions for optical pattern recognition

The projection-slice synthetic discriminant function (PSDF) filter is introduced and proposed for distortion-invariant pattern-recognition applications. The projection-slice theorem, often used in tomographic applications for medical imaging, is utilized to implement a distortion-invariant filter. Taking M projections from one training image and combining them with (N - 1)M projections taken from another N - 1 training image accomplishes this. With the projection-slice theorem, each set of these M projections can be represented as M one-dimensional slices of the two-dimensional Fourier transform of the particular training image. Therefore, the PSDF filter has the advantage of matching each of the training images with at least M slices of their respective Fourier transforms. This filter is theoretically analyzed, numerically simulated, and experimentally implemented and tested to verify the simulation results. These tests show that the PSDF filter significantly outperforms the matched-filter and the basic synthetic discriminant function technique for the particular images used.

A note on a higher-order neural network for distortion invariant pattern recognition

This paper examines capabilities of some higher-order neural networks (HONNs) for distortion invariant pattern recognition. Recently, it is shown that a single-layer HONN is effective for scale, rotation and shift invariance. Further in the training process it requires only one example for one category and a small number of iterations. However, there are problems that scale invariance doesn't hold precisely and it is not so effective for distorted unknown patterns. This paper presents some ideas to improve the recognition accuracy for scaled and distorted patterns, by changing the structure of HONN. The experimental results are presented to examine the feasibility of our approach.

Intensity- and distortion-invariant pattern recognition with complex linear morphology

A unified model based pattern recognition approach is introduced which can be formulated into a variety of techniques to be used for a variety of applications. In this approach, complex phasor addition and cancellation are incorporated into the design of filter(s) to perform implicit logical operations using linear correlation operators. These implicit logical operations are suitable to implement high level gray scale morphological transformations of input images. In this way non-linear decision boundaries are effectively projected into the input signal space yet the mathematical simplicity of linear filter designs is maintained. This approach is applied to the automatic distortion- and intensity-invariant object recognition problem. A set of shape operators or complex filters is introduced which are logically structured into a filter bank architecture to accomplish the distortion and intensity-invariant system. This synthesized complex filter bank is optimally sensitive to fractal noise representing natural scenery. The sensitivity is optimized for a specific fractal parameter range using the Fisher discriminant. The output responses of the proposed system are shown for target, clutter, and pseudo-target inputs to represent its discrimination and generalization capability in the presence of distortion and intensity variations. Its performance is demonstrated with realistic scenery as well as synthesized inputs.

USE OF GRADATED PATTERNS IN AN ASSOCIATIVE NEURAL MEMORY FOR INVARIANT PATTERN RECOGNITION

Distortion invariant pattern recognition is an interesting problem from the biological and technological point of view. However, it has not yet been solved by neural networks in satisfactory way. This paper investigates an associative neural network system to improve the recalling accuracy for distortion patterns. On a perception type of neural network with feedback, error back-propagation algorithm and energy function are used for a learning process and a recalling process, respectively. By using gradated patterns as learning and unknown patterns, it is shown that the recalling accuracy becomes higher than using original pattern themselves.

A higher-order neural network for distortion invariant pattern recognition

Recently, it is shown that a single layer, higher-order neural network is effective for scale, rotation and shift invariance and in the training process it requires only one example for one category and a very small number of iterations. However, there are problems that scale invariance doesn't hold precisely and it is not so effective for distortion of unknown patterns. In this paper we present an idea to realize the scale invariance precisely and suggest a method that is available to distorted patterns. The experimental results are presented to show the feasibility of our approach.

Advanced distortion-invariant minimum average correlation energy (MACE) filters

The original minimum average correlation energy (MACE) filter is addressed by using a new database (strategic relocatable objects, missile launchers) and including noise performance, depression angle, and resolution effects on the number of training set images that are required. Major attention is given to our new MACE filter algorithms for distortion-invariant pattern recognition: shifted-MACE filters (to suppress large false correlation peaks), minimum variance-MACE filters (for improved noise performance), multiple symbolic encoded filters (to reduce the effect of false correlation peaks), and Gaussian-MACE filters (to improve noise performance and intraclass recognition and reduce the training set size).