Real-world Imagery Research Articles

A new possibilistic clustering algorithm for line detection in real world imagery

Though promising in nature, line detection algorithms based on fuzzy clustering suffer from excessive sensitivity to noise and non-linear structures. A new detection scheme is proposed here which is suitable for the processing of real-world images. Possibilistic clustering is used instead of fuzzy clustering to achieve a higher immunity to noise, whereas a set of criteria to eliminate non-linear clusters is provided to take into account the presence of curved lines. Merging of segments is possible due to a fuzzy reasoning module exploiting human perception considerations. The number of parameters to be set is kept to a minimum, thus ensuring generality and robustness. Tests confirm the ability of the proposed system in interpreting the linear structures present in the image.

Three-dimensional target recognition and tracking using neural networks trained on optimal views

We describe a general approach to the representation and recognition of 3-D objects as it applies to automatic target recognition (ATR) tasks. The method is based on locally adaptive target segmentation, neural network classifier design, and a novel view selection mechanism that develops ‘‘visual filters’’ responsive to specific target classes that encode the complete viewing sphere with a small number of prototypical examples. The optimal set of visual filters is found via a crossvalidation-like data reduction algorithm used to train banks of backpropagation (BP) neural networks. To improve recognition accuracy under noisy or occluded conditions, as well as to eliminate false alarms, the proposed recognition system employs a temporal evidence integration technique that enables tracking and lock-on even when both targets and camera move. Experimental results on synthetic and real-world imagery demonstrate the feasibility of our approach.

Cluster validation for unsupervised stochastic model-based image segmentation

Image segmentation is an important and early processing stage in many image analysis problems. Often, this must be done in an unsupervised fashion in that training data is not available and the class-conditioned feature vectors must be estimated directly from the data. A major problem in such applications is the determination of the number of classes actually present in an image. This problem, called the cluster validation problem, remains essentially unsolved. We investigate the cluster validation problem associated with the use of a previously developed unsupervised segmentation algorithm based upon the expectation-maximization (EM) algorithm. More specifically, we consider several well-known information-theoretic criteria (ICs) as candidate solutions to the validation problem when used in conjunction with this EM-based segmentation scheme. We show that these criteria generally provide inappropriate solutions due to the domination of the penalty term by the associated log-likelihood function. As an alternative we propose a model-fitting technique in which the complete data log-likelihood functional is modeled as an exponential function in the number of classes acting. The estimated number of classes are then determined in a manner similar to finding the rise time of the exponential function. This new validation technique is shown to be robust and outperform the ICs in our experiments. Experimental results for both synthetic and real world imagery are detailed.

Fractals and image representation

For some years now, ‘fractals’ have provided spectacular displays (which have seemingly hypnotised both the popular technical press and the ‘media’) on computer workstations equipped with high-power graphics processors, and also provoked much controversy with regard to their potential for mimicking real-world imagery. In this paper the authors briefly describe the theoretical background of fractal representations and investigate in more detail the validity of such an approach for applications in natural scene portrayal and coding.

Performance of Block Cosine Image Coding with Adaptive Quantization

Quantizers for block transform image coding systems are typically designed under the assumption of Gaussian statistics for the transform coefficients. While convincing arguments can be provided in support of this approach, empirical evidence is presented demonstrating that, except possibly for the dc term, wide departures from Gaussian behavior can be expected for real-world imagery at typical block sizes. In this paper we describe the performance of a block cosine image coding system with an adaptive quantizer matched to the statistics of the transform coefficients. The adaptive quantizer is based upon a recently developed algorithm which employs a training sequence in the design procedure. At encoding rates of approximately 1 bit/pixel and above, this approach results in significant improvement in reconstructed image quality compared to fixed quantization schemes designed under the Gaussian assumption. For rates much below 1 bit/pixel the relative improvement is negligible.

Passive Visible To Infrared Transducer For Dynamic Infrared Image Simulation

Abstract. A passive visible to infrared image transducer is described which, for the first time, allows practical simulation of high resolution dynamic infrared imagery. A theoretical performance analysis is presented which indicates that an MTF (modulation transfer function) of 50% at 5 cycles/mm is achievable with a time response of less than 20 msec and a required input power of less than 2 mW/cm2 per simulated degree AT. Experimental results on working devices are reported which match theoretical predictions. Successful simulation of high speed real-world infrared imagery is described. Suggestions for future work which should further enhance resolution and dynamic range are presented.

Infrared thermography: a graphic interpretation of exitance as a function of surface temperature, emittance, and background; errata

Calorimetrio and radiation balance equations for an object in thermal equilibrium with its environment are combined to yield a graphic interpretation of exitance as a function of surface temperature, emittance, and background factors. In a monochromatic image of a scene, errors in interpreting the same shade-of-gray as being isothermal are immediately apparent from the graph. Relative effects on exitance of fractional changes in the experimental variables, internal power dissipation through the surface and emittance of the surface are derived and illustrated graphically. Modifications of the simple model move toward real world imagery.

Infrared thermography: a graphic interpretation of exitance as a function of surface temperature, emittance, and background

Calorimetrio and radiation balance equations for an object in thermal equilibrium with its environment are combined to yield a graphic interpretation of exitance as a function of surface temperature, emittance, and background factors. In a monochromatic image of a scene, errors in interpreting the same shade-of-gray as being isothermal are immediately apparent from the graph. Relative effects on exitance of fractional changes in the experimental variables, internal power dissipation through the surface and emittance of the surface are derived and illustrated graphically. Modifications of the simple model move toward real world imagery.

Robust Two-Dimensional Tree Encoding of Images

A class of robust two-dimensional (2-D) tree coding procedures are described for efficient encoding of monochrome images. The encoding filter is designed on the basis of ensemble properties of a particular class of stochastic random fields which have been shown to provide a useful model for typical imagery data. We make use of the ( M; K', L' ) algorithm in instrumenting the 2-D tree search. Results are demonstrated for both synthetic images and a number of head-and-shoulders images. These results demonstrate the effectiveness of 2-D tree encoding using fairly moderate search intensities and the inherent robustness of 2-D tree encoding more generally. In particular, it is demonstrated that, by logical design of the encoding filter under an appropriate stochastic modeling assumption, the 2-D tree encoder provides relatively efficient performance for a rather wide range of real-world imagery.

Texture Discrimination Based Upon an Assumed Stochastic Texture Model

A new approach to texture discrimination is described. This approach is based upon an assumed stochastic model for texture in imagery and is an approximation to the statistically optimum maximum likelihood classifier. The construction and properties of the stochastic texture model are described and a digital filtering implementation of the resulting maximum likelihood texture discriminant is provided. The efficacy of this approach is demonstrated through experimental results obtained with simulated texture data. A comparison is provided with more conventional texture discriminants under identical conditions. The implications to texture discrimination in realworld imagery are discussed.

Two-Dimensional DPCM Image Coding Based on an Assumed Stochastic Image Model

A new approach to the design of the feedback predictor in a two-dimensional (2-D) DPCM encoder is presented. This approach is based upon an assumed stochastic image model which exhibits the predominant and pronounced edge structure typical of real-world imagery and is in distinction from the conventional approach based upon 2-D autoregressive (AR) modeling assumptions. The performance of the resulting DPCM encoder is shown to offer significant performance advantages over existing DPCM design approaches at transmission rates of 2 bits/pixel or less. Furthermore, this new approach results in improved ability to cope with channel errors.

Stochastic image models generated by random tessellations of the plane

A useful class of two-dimensional (2-D) random fields is described which can be generated by random tessellations of the plane. The random tessellations are in turn generated by marked point processes evolving according to a spatial parameter. Gray levels are assigned within elementary disjoint regions generated by the tessellations to have specified correlation properties with gray levels in contiguous regions. A complete second-order statistical description of the resulting class of random fields is provided. This includes not only autocorrelation functions and power spectral densities but joint probability density functions. Several applications are discussed including the modeling of real-world imagery possessing inherent edge structure.