International Journal of Pattern Recognition and Artificial Intelligence

The recognition of patterns is an important task in robot and computer vision. The patterns themselves could be one- or two-dimensional, depending upon the application. Pattern matching is a computationally intensive and time consuming operation. The design of special purpose hardware could speed up the matching task considerably, making real-time responses possible. Advances in parallel processing and VLSI technologies have made it possible to implement inexpensive, efficient and very fast custom designs. Many approaches and solutions have been proposed in the literature for hardware implementations of pattern matching techniques. In this paper, we present a detailed overview of some of the important contributions in the area of hardware algorithms and architectures for pattern matching.

The use of autoregressive models in textual analysis holds great potential. Coupling the technique to a circular neighbourhood set imparts a rotational invariant property to it. This was demonstrated by Kashyap and Khotanzad in their model called the Circular Symmetric Autogressive (CSAR) Random Field model. The short-coming in this very ingenious proposal is that it is set in a background of square pixels and the rotational invariant property of the model fails in cases when the aspect ratio of the pixels are not at unity. This paper proposes a major modification to the CSAR to render the model rotational invariant under all configurations of pixel implementation. It is based on the area segments covered by a circle set in a 3×3 neighbourhood. We call it the Circular Area Autoregressive (CAAR) model. The results obtained from the CAAR showed much better consistency over that of the CSAR when a non-square pixel image was used.

In this paper, we propose a simple, but efficient method to recognize two-dimensional shapes without regard to their translation, rotation, and scaling factors. In our scheme, we use all of the boundary points to calculate the first principal component, which is the first shape feature. Next, by dividing the boundary points into groups by projecting them onto the first principal component, each shape is partitioned into several blocks. These blocks are processed separately to produce the remaining shape features. In shape matching, we compare two shapes by calculating the difference between the two sets of features to see whether the two shapes are similar or not.

The amount of storage used to represent a shape in our method is fixed, unlike most other shape recognition schemes. The time complexity of our shape matching algorithm is also O(n), where n is the number of blocks. Therefore, the matching algorithm takes little computation time, and is independent of translation, rotation, and scaling of shapes.

In this paper we address the camera parameter estimation problem using an arbitrary planar calibration object. We intend not to use a system of linear or nonlinear equations in terms of the camera parameters derived from the correspondence relationships between the image features (points or lines) and the object features. Instead, we assume that the shape of the planar object can be arbitrary. Therefore, there may not exist any extractable object point or line features. We formulate the estimation method as a generate-and-test method. To make the method effective, we reduce the dimensionality of the parameter space from six to two using a simplified model of the perspective transformation under some mild conditions. Furthermore, several other measures are also taken to cut down the generation time and the verification time. Real, computer-generated, and noisy imagery data are used to illustrate our method in the experiments. The statistics on the speed performance of the method are also included.

This paper reviews some gradient edge detection methods and proposes a new detector — the template matching edge detector (TMED). This detector utilizes the concepts of pattern analysis and the template matching of 3×3 masks. A set of performance criteria was used to evaluate the gradient edge detectors as well as the template matching edge detector. The results indicate that the new method is superior to the other gradient edge detectors. In addition, the template matching edge detector has also demonstrated good performance on noisy images. It can obtain very precise edge detection of single pixel width.

The 2D string, proposed by Chang et al., is a spatial index structure which preserves the information of spatial relationships in a spatial database. In this paper, two new image retrieval algorithms for 2D string are proposed. The first one improves the retrieval efficiency, while the second reduces the space requirement. The performance analysis shows that the two methods perform much better than previous works especially when the spatial database is large.

The paper describes the design and implementation of an information retrieval system using color as the index in its color image archive. Salient aspects of this approach are the use of direct visual representation of color in querying and of fuzzy set theory to formally represent intrinsic human uncertainty in evaluating the similarity between colors (query interpretation).

The steps in which the information retrieval strategy is organized are illustrated, and an example of its application given, showing how the implemented system can be tailored to manage archives of images representing fabric samples.

This paper presents a hybrid approach towards accomplishing the task of range image segmentation. It tends to combine region growing and one-dimensional orthogonal polynomial fitting techniques for the detection of edges in range images. Edges detected by such a method possesses good localisation property. This aids to steer the region growing process towards accomplishing accurate border partitioning. In addition, the incorporation of region growing process eliminates internal micro edges and provides for missing border reconstruction as it is able to detect weak edges. It is believed that the edge and segmentation maps produced, when conveyed to the higher level recognition process, may prove valuable not only for CAD based modeling purposes but may also aid to provide descriptive syntax to the object identification process.