Showing papers on "Feature extraction published in 1994"

Good features to track

Abstract: No feature-based vision system can work unless good features can be identified and tracked from frame to frame. Although tracking itself is by and large a solved problem, selecting features that can be tracked well and correspond to physical points in the world is still hard. We propose a feature selection criterion that is optimal by construction because it is based on how the tracker works, and a feature monitoring method that can detect occlusions, disocclusions, and features that do not correspond to points in the world. These methods are based on a new tracking algorithm that extends previous Newton-Raphson style search methods to work under affine image transformations. We test performance with several simulations and experiments. >

Using mutual information for selecting features in supervised neural net learning

Abstract: This paper investigates the application of the mutual information criterion to evaluate a set of candidate features and to select an informative subset to be used as input data for a neural network classifier. Because the mutual information measures arbitrary dependencies between random variables, it is suitable for assessing the "information content" of features in complex classification tasks, where methods bases on linear relations (like the correlation) are prone to mistakes. The fact that the mutual information is independent of the coordinates chosen permits a robust estimation. Nonetheless, the use of the mutual information for tasks characterized by high input dimensionality requires suitable approximations because of the prohibitive demands on computation and samples. An algorithm is proposed that is based on a "greedy" selection of the features and that takes both the mutual information with respect to the output class and with respect to the already-selected features into account. Finally the results of a series of experiments are discussed. >

Image Processing

Abstract: MUCKE aims to mine a large volume of images, to structure them conceptually and to use this conceptual structuring in order to improve large-scale image retrieval. The last decade witnessed important progress concerning low-level image representations. However, there are a number problems which need to be solved in order to unleash the full potential of image mining in applications. The central problem with low-level representations is the mismatch between them and the human interpretation of image content. This problem can be instantiated, for instance, by the incapability of existing descriptors to capture spatial relationships between the concepts represented or by their incapability to convey an explanation of why two images are similar in a content-based image retrieval framework. We start by assessing existing local descriptors for image classification and by proposing to use co-occurrence matrices to better capture spatial relationships in images. The main focus in MUCKE is on cleaning large scale Web image corpora and on proposing image representations which are closer to the human interpretation of images. Consequently, we introduce methods which tackle these two problems and compare results to state of the art methods. Note: some aspects of this deliverable are withheld at this time as they are pending review. Please contact the authors for a preview.

View-based and modular eigenspaces for face recognition

Abstract: We describe experiments with eigenfaces for recognition and interactive search in a large-scale face database. Accurate visual recognition is demonstrated using a database of O(10/sup 3/) faces. The problem of recognition under general viewing orientation is also examined. A view-based multiple-observer eigenspace technique is proposed for use in face recognition under variable pose. In addition, a modular eigenspace description technique is used which incorporates salient features such as the eyes, nose and mouth, in an eigenfeature layer. This modular representation yields higher recognition rates as well as a more robust framework for face recognition. An automatic feature extraction technique using feature eigentemplates is also demonstrated. >

Transform features for texture classification and discrimination in large image databases

Abstract: Proposes a method for classification and discrimination of textures based on the energies of image subbands. The authors show that with this relatively simple feature set, effective texture discrimination can be achieved. In the paper, subband-energy feature sets extracted from the following typical image decompositions are compared: wavelet subband, uniform subband, discrete cosine transform (DCT), and spatial partitioning. The authors report that over 90% correct classification was attained using the feature set in classifying the full Brodatz [1965] collection of 112 textures. Furthermore, the subband energy-based feature set can be readily applied to a system for indexing images by texture content in image databases, since the features can be extracted directly from spatial-frequency decomposed image data. The authors also show that to construct a suitable space for discrimination, Fisher discrimination analysis (Dillon and Goldstein, 1984) can be used to compact the original features into a set of uncorrelated linear discriminant functions. This procedure makes it easier to perform texture-based searches in a database by reducing the dimensionality of the discriminant space. The authors also examine the effects of varying training class size, the number of training classes, the dimension of the discriminant space and number of energy measures used for classification. The authors hope that the performance for texture discrimination of these simple energy-based features will allow images in a database to be efficiently and effectively indexed by contents of their textured regions. >

Mammographic feature enhancement by multiscale analysis

Abstract: Introduces a novel approach for accomplishing mammographic feature analysis by overcomplete multiresolution representations. The authors show that efficient representations may be identified within a continuum of scale-space and used to enhance features of importance to mammography. Methods of contrast enhancement are described based on three overcomplete multiscale representations: 1) the dyadic wavelet transform (separable), 2) the /spl phi/-transform (nonseparable, nonorthogonal), and 3) the hexagonal wavelet transform (nonseparable). Multiscale edges identified within distinct levels of transform space provide local support for image enhancement. Mammograms are reconstructed from wavelet coefficients modified at one or more levels by local and global nonlinear operators. In each case, edges and gain parameters are identified adaptively by a measure of energy within each level of scale-space. The authors show quantitatively that transform coefficients, modified by adaptive nonlinear operators, can make more obvious unseen or barely seen features of mammography without requiring additional radiation. The authors' results are compared with traditional image enhancement techniques by measuring the local contrast of known mammographic features. They demonstrate that features extracted from multiresolution representations can provide an adaptive mechanism for accomplishing local contrast enhancement. By improving the visualization of breast pathology, one can improve chances of early detection while requiring less time to evaluate mammograms for most patients. >

A Framework for Low Level Feature Extraction

Abstract: The paper presents a framework for extracting low level features. Its main goal is to explicitely exploit the information content of the image as far as possible. This leads to new techniques for deriving image parameters, to either the elimination or the elucidation of ”buttons”, like thresholds, and to interpretable quality measures for the results, which may be used in subsequent steps. Feature extraction is based on local statistics of the image function. Methods are available for blind estimation of a signal dependent noise variance, for feature preserving restoration, for feature detection and classification, and for the location of general edges and points. Their favorable scale space properties are discussed.

Signal processing and compression with wavelet packets

Abstract: Wavelet packets are a versatile collection of functions generalizing the compactly supported wavelets of Daubechies. They are used to analyze and manipulate signals such as sound and images. We describe a library of such waveforms and demonstrate a few of their analytic properties. We also describe an algorithm to chose a best basis subset, tailored to fit a specific signal or class of signals. We apply this algorithm to two signal processing tasks: acoustic signal compression, and feature extraction in certain images.

Handwriting signal processing front-end for handwriting recognizers

Abstract: A handwriting signal processing front-end method and apparatus for a handwriting training and recognition system which includes non-uniform segmentation and feature extraction in combination with multiple vector quantization. In a training phase, digitized handwriting samples are partitioned into segments of unequal length. Features are extracted from the segments and are grouped to form feature vectors for each segment. Groups of adjacent from feature vectors are then combined to form input frames. Feature-specific vectors are formed by grouping features of the same type from each of the feature vectors within a frame. Multiple vector quantization is then performed on each feature-specific vector to statistically model the distributions of the vectors for each feature by identifying clusters of the vectors and determining the mean locations of the vectors in the clusters. Each mean location is represented by a codebook symbol and this information is stored in a codebook for each feature. These codebooks are then used to train a recognition system. In the testing phase, where the recognition system is to identify handwriting, digitized test handwriting is first processed as in the training phase to generate feature-specific vectors from input frames. Multiple vector quantization is then performed on each feature-specific vector to represent the feature-specific vector using the codebook symbols that were generated for that feature during training. The resulting series of codebook symbols effects a reduced representation of the sampled handwriting data and is used for subsequent handwriting recognition.

Geometric primitive extraction using a genetic algorithm

Abstract: Extracting geometric primitives from geometric sensor data is an important problem in model-based vision. A minimal subset is the smallest number of points necessary to define a unique instance of a geometric primitive. A genetic algorithm based on a minimal subset representation is used to perform primitive extraction. It is shown that the genetic approach is an improvement over random search and is capable of extracting more complex primitives than the Hough transform. >

Visualizing features and tracking their evolution

Abstract: We describe basic algorithms to extract coherent amorphous regions (features or objects) from 2 and 3D scalar and vector fields and then track them in a series of consecutive time steps. We use a combination of techniques from computer vision, image processing, computer graphics, and computational geometry and apply them to data sets from computational fluid dynamics. We demonstrate how these techniques can reduce visual clutter and provide the first step to quantifying observable phenomena. These results can be generalized to other disciplines with continuous time-dependent scalar (and vector) fields. >

Active mesh-a feature seeking and tracking image sequence representation scheme

Abstract: This paper introduces a representation scheme for image sequences using nonuniform samples embedded in a deformable mesh structure. It describes a sequence by nodal positions and colors in a starting frame, followed by nodal displacements in the following frames. The nodal points in the mesh are more densely distributed in regions containing interesting features such as edges and corners; and are dynamically updated to follow the same features in successive frames. They are determined automatically by maximizing feature (e.g., gradient) magnitudes at nodal points, while minimizing interpolation errors within individual elements, and matching errors between corresponding elements. In order to avoid the mesh elements becoming overly deformed, a penalty term is also incorporated, which measures the irregularity of the mesh structure. The notions of shape functions and master elements commonly used in the finite element method have been applied to simplify the numerical calculation of the energy functions and their gradients. The proposed representation is motivated by the active contour or snake model proposed by Kass, Witkin, and Terzopoulos (1988). The current representation retains the salient merit of the original model as a feature tracker based on local and collective information, while facilitating more accurate image interpolation and prediction. Our computer simulations have shown that the proposed scheme can successfully track facial feature movements in head-and-shoulder type of sequences, and more generally, interframe changes that can be modeled as elastic deformation. The treatment for the starting frame also constitutes an efficient representation of arbitrary still images. >

Real-time 3-D pose estimation using a high-speed range sensor

Abstract: This paper describes a system which can perform full 3-D pose estimation of a single arbitrarily shaped, rigid object at rates up to 10 Hz. A triangular mesh model of the object to be tracked is generated offline using conventional range sensors. Real-time range data of the object is sensed by the CMU high speed VLSI range sensor. Pose estimation is performed by registering the real-time range data to the triangular mesh model using an enhanced implementation of the Iterative Closest Point (ICP) Algorithm introduced by Besl and McKay (1992). The method does not require explicit feature extraction or specification of correspondence. Pose estimation accuracies of the order of 1% of the object size in translation, and 1 degree in rotation have been measured. >

Extracting road curvature and orientation from image edge points without perceptual grouping into features

Abstract: The ARCADE (Automated Road Curvature And Direction Estimation) algorithm estimates road curvature and tangential road orientation relative to the camera line-of-sight. The input to ARCADE consists of edge point locations and orientations extracted from an image, and it performs the estimation without the need for any prior perceptual grouping of the edge points into individual lane boundaries. It is able to achieve this through the use of global constraints on the individual lane boundary shapes derived from an explicit model of road structure in the world. The use of the least median squares robust estimation technique allows the algorithm to function correctly in cases where up to 50% of the input edge data points are contaminating noise. Two applications of ARCADE as the first stage of processing for a lane sensing task are described: 1) the extraction of the locations of the features defining the visible lane structure of the road; and 2) the generation of training instances for an ALVINN-like neural network road follower.

Local discriminant bases

Abstract: We describe an extension to the `best-basis' method to construct an orthonormal basis which maximizes a class separability for signal classification problems This algorithm reduces the dimensionality of these problems by using basis functions which are well localized in time- frequency plane as feature extractors We tested our method using two synthetic datasets: extracted features (expansion coefficients of input signals in these basis functions), supplied them to the conventional pattern classifiers, then computed the misclassification rates These examples show the superiority of our method over the direct application of these classifiers on the input signals As a further application, we also describe a method to extract signal component from data consisting of signal and textured background

Recovery of ego-motion using image stabilization

Abstract: A method for computing the 3D camera motion (the ego-motion) in a static scene is introduced, which is based on computing the 2D image motion of a single image region directly from image intensities. The computed image motion of this image region is used to register the images so that the detected image region appears stationary. The resulting displacement field for the entire scene between the registered frames is affected only by the 3D translation of the camera. After canceling the effects of the camera rotation by using such 2D image registration, the 3D camera translation is computed by finding the focus-of-expansion in the translation-only set of registered frames. This step is followed by computing the camera rotation to complete the computation of the ego-motion. The presented method avoids the inherent problems in the computation of optical flow and of feature matching, and does not assume any prior feature detection or feature correspondence. >

Gray-scale morphological granulometric texture classification

Abstract: Binary morphological granulometric size distributions were conceived by Matheron as a way of describing image granularity (or texture). Since each normalized size distribution is a probability density, feature vectors of granulometric moments result. Recent application has focused on taking local size distributions around individual pixels so that the latter can be classified by surrounding texture. The extension of the local-classification technique to gray-scale textures is investigated. It does so by using 42 granulometric features, half generated by opening granulometries and a dual half generated by closing granulometries. After training and classification of both dependent and independent data, feature extraction (compression) is accomplished by means of the Karhunen-Loeve transform. Sequential feature selection is also applied. The effect of randomly placed uniform noise is investigated. In particular, the degree to which training in noise increases robustness across noise levels is studied, and feature selection is employed to arrive at a noise-insensitive set of granulometric classifiers.

On-line cursive script recognition using time delay neural networks and hidden Markov models

Abstract: Presents a writer independent system for on-line handwriting recognition which can handle both cursive script and hand-print. The pen trajectory is recorded by a touch sensitive pad, such as those used by note-pad computers. The input to the system contains the pen trajectory information, encoded as a time-ordered sequence of feature vectors. Features include X and Y coordinates, pen-lifts, speed, direction and curvature of the pen trajectory. A time delay neural network with local connections and shared weights is used to estimate a posteriori probabilities for characters in a word. A hidden Markov model segments the word into characters in a way which optimizes the global word score, taking a dictionary into account. A geometrical normalization scheme and a fast but efficient dictionary search are also presented. Trained on 20000 unconstrained cursive words from 59 writers and using a 25000 word dictionary the authors reached a 89% character and 80% word recognition rate on test data from a disjoint set of writers. >

Method for identifying objects and features in an image

Abstract: The present invention features the use of the fundamental concept of color perception and multi-level resolution to perform scene segmentation and object/feature extraction in the context of self-determining and self-calibration modes. The technique uses only a single image, instead of multiple images as the input to generate segmented images. Moreover, a flexible and arbitrary scheme is incorporated, rather than a fixed scheme of segmentation analysis. The process allows users to perform digital analysis using any appropriate means for object extraction after an image is segmented. First, an image is retrieved. The image is then transformed into at least two distinct bands. Each transformed image is then projected into a color domain or a multi-level resolution setting. A segmented image is then created from all of the transformed images. The segmented image is analyzed to identify objects. Object identification is achieved by matching a segmented region against an image library. A featureless library contains full shape, partial shape and real-world images in a dual library system. The depth contours and height-above-ground structural components constitute a dual library. Also provided is a mathematical model called a Parzen window-based statistical/neural network classifier, which forms an integral part of this featureless dual library object identification system. All images are considered three-dimensional. Laser radar based 3-D images represent a special case.

Learning and feature selection in stereo matching

Abstract: We present a novel stereo matching algorithm which integrates learning, feature selection, and surface reconstruction. First, a new instance based learning (IBL) algorithm is used to generate an approximation to the optimal feature set for matching. In addition, the importance of two separate kinds of knowledge, image dependent knowledge and image independent knowledge, is discussed. Second, we develop an adaptive method for refining the feature set. This adaptive method analyzes the feature error to locate areas of the image that would lead to false matches. Then these areas are used to guide the search through feature space towards maximizing the class separation distance between the correct match and the false matches. Third, we introduce a self-diagnostic method for determining when apriori knowledge is necessary for finding the correct match. If the a priori knowledge is necessary then we use a surface reconstruction model to discriminate between match possibilities. Our algorithm is comprehensively tested against fixed feature set algorithms and against a traditional pyramid algorithm. Finally, we present and discuss extensive empirical results of our algorithm based on a large set of real images. >

Feature matching for building extraction from multiple views

Abstract: We present a method to generate three-dimensional building hypotheses starting from sparse features, hypothesized building corners, extracted from multiple views of the scene. This technique relies on knowledge about the imaging geometry and acquisition parameters to provide rigorous geometric constraints for the matching process. The effectiveness of this approach is demonstrated using complex aerial imagery taken with highly oblique views containing buildings with flat or peaked roofs. >

Task driven perceptual organization for extraction of rooftop polygons

Abstract: A new method for extracting planar polygonal rooftops in monocular aerial imagery is proposed. Structural features are extracted and hierarchically related using perceptual grouping techniques. Top-down feature verification is used so that features, and links between the features, are verified with local information in the image and weighed in a graph. Cycles in the graph correspond to possible building rooftop hypotheses. Virtual features are hypothesized for the perceptual completion of partially occluded rooftops. Extraction of the "best" grouping features into a building rooftop hypothesis is posed as a graph search problem. The maximally weighted, independent set of cycles in the graph is extracted as the final set of roof boundaries. >

Edge extraction by FIRE operators

Abstract: FIRE (fuzzy inference ruled by else-action) operators are a recently proposed family of fuzzy operators for image processing. After an introduction of the generalized structure of the FIRE edge extractor, in this paper it is shown how FIRE operators can be designed in order to comply with the following two requirements: 1) extraction of edges from a noiseless image by means of the simplest possible rule-base; 2) extraction of edges from a noisy image. Some experimental results show the performances of the proposed approach. >

A genetic algorithm for the detection of 2D geometric primitives in images

Abstract: We investigate the use of genetic algorithms (GAs) for image primitives extraction (such as segments, circles, ellipses or quadrilaterals). This approach completes the well-known Hough transform, in the sense that GAs are efficient when the Hough approach becomes too expensive in memory, i.e. when we search for complex primitives having more than 3 or 4 parameters. A GA is a stochastic technique, relatively slow, but which provides with an efficient tool to search in a high dimensional space. The philosophy of the method is very similar to the Hough transform, which is to search an optimum in a parameter space. However, we will see that the implementation is different.

A system for discovering relationships by feature extraction from text databases

Abstract: A method for accessing text-based information using domain-specific features rather than documents alone is presented. The basis of this approach is the ability to automatically extract features from large text databases, and identify statistically significant relationships or associations between those features. The techniques supporting this approach are discussed, and examples from an application using these techniques, named the Associations System, are illustrated using the Wall Street Journal database. In this particular application, the features extracted are company and person names. The series of tests run on the Associations System demonstrate that feature extraction can be quite accurate, and that the relationships generated are reliable. In addition to conventional measures of recall and precision, evaluation measures are currently being studied which will indicate the usefulness of the relationships identified, in various domain-specific contexts.

Restructured recursive DCT and DST algorithms

Abstract: The discrete cosine transform (DCT) and the discrete sine transform (DST) have found wide applications in speech and image processing, as well as telecommunication signal processing for the purpose of data compression, feature extraction, image reconstruction, and filtering. In this paper, we present new recursive algorithms for the DCT and the DST. The proposed method is based on certain recursive properties of the DCT coefficient matrix, and can be generalized to design recursive algorithms for the 2-D DCT and the 2-D DST. These new structured recursive algorithms are able to decompose the DCT and the DST into two balanced lower-order subproblems in comparison to previous research works. Therefore, when converting our algorithms into hardware implementations, we require fewer hardware components than other recursive algorithms. Finally, we propose two parallel algorithms for accelerating the computation. >

Image processing system and method thereof in which three dimensional shape is reproduced from two dimensional image data

Abstract: The image processing system has a unit for photographing an object in two dimensions, a feature extraction unit for extracting features from the two-dimensional image data from the photographing means, and a three-dimensional shape reproduction unit. The feature extraction unit refers to feature points given to the object to extract the features. The three-dimensional shape reproduction unit expresses the object by a dynamic equation, applies force from the feature extraction coordinates to the dynamic model to cause the dynamic model to change shape and supplement depth data, and to thereby reproduce the three-dimensional shape of the object. To increase the speed of the processing, it is desirable to divide the image data of the object into portions with little changes in shape and perform the processing for reproducing the three-dimensional shape for each mode.