Jaroslaw Szostakowski

Mixed neural-traditional classifier for character recognition

In this paper, we present the efficient voting classifier for the recognition of handwritten and printed characters. This system consists of three voting nonlinear classifiers: two of them based on the multilayer perceptron, and one uses the moments method. The combination of these kinds of systems shows superiority of neural techniques applied with classical against exclusive traditional approach and results in high percentage of correctly recognized characters. Also, we present a comparison of the recognition results.

Image feature extraction using mathematical morphology

Mathematical morphology (MM) is a very efficient tool for image processing, based on non- linear operators.In this paper MM is applied to extract the images features. As a feature we understand specific information about the image i.e. location, size, orientation of certain image elements. Morphological operators are applied to find and measure objects on the images surface. Two practical examples are considered. First is devoted to analysis of binary images, containing printed characters. Characters are separated and MM is used to extract some information from each character. These features are later measured and included in a feature vector. It contains the special kind of information - the number of elements of the character with its shape modified in different ways. Second examples shows how feature extraction by MM works on graytone images. Images for analysis contain human faces. Morphological operators extract some important elements of human face. This information is very important to identify the human face. Experiments show us how the morphological operators can be applied to the feature detection. The simplest operators as erosion and dilation, as well as more sophisticated tools like: morphological filtering, geodesic transformations are used for that purpose. Also directional operations are applied to extract some areas. This paper includes algorithms for feature extraction by MM, as well as the brief description of morphological tools, explication of experiments and the results of them.

Neural implementation of ARMA type filters for image restoration

We present a novel neural implementation of the autoregressive moving average (ARMA) type filters for image deblurring. Our filter is designed on the basis of a known blur system. As the neural net, we used a multilayer perceptron. Due to connection of the parallel processing and nonlinear characteristics in the neural networks, we hoped to reduced the influence of noise and roundoff errors. We present the construction of different learning patterns for this net. Some practical examples are shown.

Classical and neural methods of image sequence interpolation

An image interpolation problem is often encountered in many areas. Some examples are interpolation for coding/decoding process for transmission purposes, reconstruction a full frame from two interlaced sub-frames in normal TV or HDTV, or reconstruction of missing frames in old destroyed cinematic sequences. In this paper an overview of interframe interpolation methods is presented. Both direct as well as motion compensated interpolation techniques are given by examples. The used methodology can also be either classical or based on neural networks depending on demand of a specific interpolation problem solving person.

Neural least-squares image filter with positive constraints

In this paper a novel neural approach for restoration of grey images degraded by known blur function and additive noise is presented. The neuron-like optimizer solver is used for constrained least-squares image filtering with positive constraints. The solution is based on penalty function methods for nonlinear, optimization problems. Continuous and discrete filter realizations are described. The practical examples is given.

Mathematical morphology and higher-order neural networks

Mathematical morphology (MM) is one of the most efficient tools in advanced digital image processing. Morphological techniques have been successfully applied in such cases as: image analysis, smoothing, enhancement, edge detection, skeletonization, filtering, and segmentation (watershed algorithms). Two essential operations of MM are dilation and erosion and can be implemented in several different ways. In our paper we propose their effective implementation by using higher order neural network approach (functional-link network). The novel structure and its learning method is presented. Some other neural network methods for MM operations are shown and compared with our approach.

Advanced Neural Networks Methods for Recognition of Handwritten Characters

In this paper, we present the efficient voting classifier for the recognition of handwritten characters. This system consists of three voting nonlinear classifiers: two of them base on the multilayer perceptron, and one uses the moments method. The combination of these kinds of systems showed superiority of neural techniques applied with classical against exclusive traditional approach and resulted in high percentage of correctly recognized characters. Also, we present a comparison of the recognition results.

Efficient filtering of color image sequences

The first step in image filtering were filters developed on the simply binary images. That was caused because early filters were strictly limited by computational and memory resources of first computers. During the time most algorithms were applied to filter gray-scale images. Advances in high integrated microprocessor technology make possible to work on higher resolution images with more important details. Image processing has been developed very fast based on still 2-D gray-scale images. We could find most of up to now developed algorithms in many monographs.

Motion compensated neural filters for video noise reduction

Time-sequential imagery can be acquired by film-based motion camera or electronic video cameras. In this case, there are several factors related to imaging sensor limitations that contribute to the graininess of resulting images. Further, in the case of image sequence compression, random noise increases the entropy of the image sequence and therefore hinders effective compression. Thus, filtering of time- sequential imagery for noise suppression is often a desirable preprocessing step. Some of video image filtering methods use the information about motion in video for reduction of noise. The most of them are based on 3D median or average filters, which supports are along motion trajectories. In this approach, it is difficult to design the proper structure of the 3D filter by analytic methods. The artificial neural networks can be useful tool for creating the structures of the filters. In this paper the novel neural networks approach to motion compensated temporal and spatio-temporal filtering is proposed. The multilayer perceptrons and functional-link nets are used for the 3D filtering. The spatio-temporal patterns are creating from real motion video images. the neural networks learn these patterns. The practical examples of the filtering are shown and compared with traditional motion-compensated filters.

Neural model-based motion estimation

In this paper a novel method of estimating displacement of moving objects from one frame to the next in the image sequence is presented. This method is based on using the artificial neural networks for different models of motion. The two model is examined: affine flow and planar surface motion. Various circuit architectures of simple neuron-like processors are considered for estimation of motion parameters. The efficiency of the proposed networks are investigated by computer simulation for using in video processing.