Heggere S. Ranganath

Perfect image segmentation using pulse coupled neural networks

This paper describes a method for segmenting digital images using pulse coupled neural networks (PCNNs). The pulse coupled neuron (PCN) model used in PCNN is a modification of Eckhorns cortical neuron model. A single layered laterally connected PCNN is capable of perfectly segmenting digital images even when there is a considerable overlap in the intensity ranges of adjacent regions. Conditions for perfect image segmentation are derived. It is also shown that addition of an inhibition receptive field to the neuron model increases the possibility of perfect segmentation. The inhibition input reduces the overlap of intensity ranges of adjacent regions by effectively compressing the intensity range of each region.

Object detection using pulse coupled neural networks

This paper describes an object detection system based on pulse coupled neural networks. The system is designed and implemented to illustrate the power, flexibility and potential the pulse coupled neural networks have in real-time image processing. In the preprocessing stage, a pulse coupled neural network suppresses noise by smoothing the input image. In the segmentation stage, a second pulse coupled neural-network iteratively segments the input image. During each iteration, with the help of a control module, the segmentation network deletes regions that do not satisfy the retention criteria from further processing and produces an improved segmentation of the retained image. In the final stage each group of connected regions that satisfies the detection criteria is identified as an instance of the object of interest.

Using association rules as texture features

A new type of texture feature based on association rules is proposed in this paper. Association rules have been used in applications such as market basket analysis to capture relationships present among items in large data sets. It is shown that association rules can be adapted to capture frequently occurring local structures in images. Association rules capture both structural and statistical information, and automatically identifies the structures that occur most frequently and relationships that have significant discriminative power. Methods for classification and segmentation of textured images using association rules as texture features are described. Simulation results using images consisting of man made and natural textures show that association rule features perform well compared to other widely used texture features. It is shown that association rule features can distinguish texture pairs with identical first, second, and third order statistics, and texture pairs that are not easily discriminable visually.

Pulse coupled neural networks for image processing

Studies of cats and monkeys visual cortex has led to the development of pulse coupled neurons which are significantly different from the conventional artificial neurons. Pulse coupled neural networks (PCNN) are modeled to capture the essence of recent understanding of image interpretation process in biological neural systems. It is shown that pulse coupled neural networks are capable of image smoothing, image segmentation and feature extraction.

Fuzzy relaxation approach for inexact scene matching

Abstract A graph theoretic approach for matching imperfectly segmented images with stored scene models is presented. The segmentation errors addressed are missing objects, extra objects, mismeasured relations, mismeasured attributes, split objects, and merged objects. By combining enhanced fuzzy relaxation and association graph techniques, the mthod integrates global inter-object relations and local object attributes to obtain more reliable matching. Problems of oversegmentation and undersegmentation are handled by modifying the association graph to include nodes involving merged regions and objects.

Iterative segmentation using pulse-coupled neural networks

Recent studies of the visual cortices of cats and monkeys has led to the development of a new class of artificial neuron models. Eckhorn and his co-workers have developed one such neuron model. They have demonstrated that the recurrent networks of Eckhorns neurons are capable of duplicating some of the neuro-physiological phenomena observed in cats visual cortex. We have modified Eckhorns neuron model in a way that the resulting neuron, referred to as the pulsed coupled neuron, becomes more suitable for image processing applications than his original model. It has been shown that a single layered laterally connected pulse coupled neural network (PCNN) is capable of smoothing, segmenting digital images. This paper describes an iterative segmentation scheme that utilizes smoothing, segmentation and feature extraction capabilities of PCNN. The knowledge driven iterative segmentation technique is powerful, flexible and has potential in real-time image processing systems.

Image segmentation using association rule features

A new type of texture feature based on association rules is described. Association rules have been used in applications such as market basket analysis to capture relationships present among items in large data sets. It is shown that association rules can be adapted to capture frequently occurring local structures in images. The frequency of occurrence of these structures can be used to characterize texture. Methods for segmentation of textured images based on association rule features are described. Simulation results using images consisting of man made and natural textures show that association rule features perform well compared to other widely used texture features. Association rule features are used to detect cumulus cloud fields in GOES satellite images and are found to achieve higher accuracy than other statistical texture features for this problem.

Techniques and Experience in Mining RemotelySensed Satellite Data

The paper presents a set of requirements for a datamining system for mining remotely sensed satellitedata based on a number of taxonomies that characterizemining of such data. The first of these taxonomies isbased on knowledge of the mining objectives and miningalgorithms. The second is based on variousrelationships that are found in data, including thosebetween different types of data, different spatiallocations of the data and different times of datacapture. The paper then describes the ADaM data miningsystem, which was developed to address theserequirements. The paper describes several data miningtechniques that have been applied to remotely senseddata. The first type is target independent mining,which mines data for transients and trends, with minedresults representing a highly concentrated form of theoriginal data. The second type is the mining ofvectors (representing multi-spectral or fused data)for association rules representing relationshipsbetween the various types of data represented by theelements of the vector. The third type mines data forassociation rules that characterize the texture of thedata.

Self partitioning neural networks for target recognition

Automatic target recognition (ATR) is a domain in which the neural network technology has been applied with limited success. The domain is characterized by large training sets with dissimilar target images carrying conflicting information. This paper presents a novel method for quantifying the degree of non-cooperation that exists among the target members of the training set. Both the network architecture and the training algorithm are considered in the computation of the non-cooperation measures. Based on these measures, the self partitioning neural network (SPNN) approach partitions the target vectors into an appropriate number of groups and trains one subnetwork to recognize the targets in each group. A fusion network combines the outputs of the subnetworks to produce the final response. This method automatically determines the number of subnetworks needed without excessive computation. The subnetworks are simple with only one hidden layer and one unit in the output layer. They are topologically identical to one another. The simulation results indicate that the method is robust and capable of self organization to overcome the ill effects of the non-cooperating targets in the training set. The self partitioning approach improves the classification accuracy and reduces the training time of neural networks significantly. It is also shown that a trained self partitioning neural network is capable of learning new training vectors without retraining on the combined training set (i.e., the training set consisting of the previous and newly acquired training vectors).

Smart adaptive optic systems using spatial light modulators

Many factors contribute to the aberrations induced in an optical system. Atmospheric turbulence between the object and the imaging system, physical or thermal perturbations in optical elements degrade the systems point spread function, and misaligned optics are the primary sources of aberrations that affect image quality. The design of a nonconventional real-time adaptive optic system using a micro-mirror device for wavefront correction is presented. The unconventional compensated imaging system presented offers advantages in speed, cost, power consumption, and weight. A pulsed-coupled neural network is used to as a preprocessor to enhance the performance of the wavefront sensor for low-light applications. Modeling results that characterize the system performance are presented.