This paper describes a new technique for object recognition based on learning appearance models. The image is decomposed into local regions which are described by a new texture representation called "Generalized Second Moments" thatare derived from the output of multiscale, multiorientation filter banks. Class-characteristic local texture features and their global composition is learned by a hierarchical mixture of experts architecture (Jordan & Jacobs). The technique is applied to a vehicle database consisting of 5 general car categories (Sedan, Van with backdoors, Van without backdoors, old Sedan, and Volkswagen Bug). This is a difficult problem with considerable in-class variation. The new technique has a 6.5% misclassification rate, compared to eigen-images which give 17.4% misclassification rate, and nearest neighbors which give 15 .7%

Curtis Padgett Department of Computer Science University of California, San Diego La Jolla, CA 92034 GarrisonCottrell Department of Computer Science University of California, San Diego La Jolla, CA 92034 Abstract We compare the generalization performance of three distinct representation schemesfor facial emotions using a single classification strategy (neural network). The face images presented to the classifiers arerepresented as: full face projections of the dataset onto their eigenvectors (eigenfaces); a similar projection constrained to eye and mouth areas (eigenfeatures); and finally a projection of the eye and mouth areas onto the eigenvectors obtained from 32x32 random image patches from the dataset. The latter system achieves 86% generalization on novel face images (individuals the networks were not trained on) drawn from a database in which human subjects consistentlyidentify a single emotion for the face . 1 Introduction Some of the most successful research in machine perception of complex natural image objects (like faces), has relied heavily on reduction strategies that encode an object as a set of values that span the principal component subspace of the object's images [Cottrell and Metcalfe, 1991, Pentland et al., 1994]. This approach has gained wide acceptance for its success in classification, for the efficiency in which the eigenvectors can be calculated, and because the technique permits an implementation thatis biologically plausible. The procedure followed in generating these face representations requires normalizing a large set of face views (" mugshots") and from these, identifying a statistically relevant subspace.

This paper describes a new technique for object recognition based on learning appearance models. The image is decomposed into local regions which are described by a new texture representation called "Generalized Second Moments" that are derived from the output of multiscale, multiorientation filter banks. Class-characteristic local texture features and their global composition is learned by a hierarchical mixture of experts architecture (Jordan & Jacobs). The technique is applied to a vehicle database consisting of 5 general car categories (Sedan, Van with backdoors, Van without backdoors, old Sedan, and Volkswagen Bug). This is a difficult problem with considerable in-class variation.

We compare the generalization performance of three distinct representation schemes for facial emotions using a single classification strategy (neural network). The face images presented to the classifiers are represented as: full face projections of the dataset onto their eigenvectors (eigenfaces); a similar projection constrained to eye and mouth areas (eigenfeatures); and finally a projection of the eye and mouth areas onto the eigenvectors obtained from 32x32 random image patches from the dataset. The latter system achieves 86% generalization on novel face images (individuals the networks were not trained on) drawn from a database in which human subjects consistently identify a single emotion for the face. 1 Introduction Some of the most successful research in machine perception of complex natural image objects (like faces), has relied heavily on reduction strategies that encode an object as a set of values that span the principal component subspace of the object's images [Cottrell and Metcalfe, 1991, Pentland et al., 1994]. This approach has gained wide acceptance for its success in classification, for the efficiency in which the eigenvectors can be calculated, and because the technique permits an implementation that is biologically plausible. The procedure followed in generating these face representations requires normalizing a large set of face views (" mugshots") and from these, identifying a statistically relevant subspace.

The National Aeronautics and Space Administration (NASA) is being challenged to perform more frequent and intensive space-exploration missions at greatly reduced cost. Nowhere is this challenge more acute than among robotic planetary exploration missions that the Jet Propulsion Laboratory (JPL) conducts for NASA. This article describes recent and ongoing work on spacecraft autonomy and ground systems that builds on a legacy of existing success at JPL applying AI techniques to challenging computational problems in planning and scheduling, real-time monitoring and control, scientific data analysis, and design automation.

To exploit their properties at a system level, we developed parallel image processing architectures for applications that rely mostly on the qualitative properties of the optical flow, rather than on the precise values of the velocity vectors. Specifically, we designed two parallel architectures that employ arrays of elementary motion sensors for the computation of heading direction and time-to-contact. The application domain that we took into consideration for the implementation of such architectures, is the promising one of vehicle navigation. Having defined the types of images to be analyzed and the types of processing to perform, we were able to use a priori infor- VLSI Architectures for Computation of Heading Direction and Time-to-contact 721 mation to integrate selectively the sparse data obtained from the velocity sensors and determine the qualitative properties of the optical flow field of interest.

To exploit their properties at a system level, we developed parallel image processing architectures for applications that rely mostly on the qualitative properties of the optical flow, rather than on the precise values of the velocity vectors. Specifically, we designed two parallel architectures that employ arrays of elementary motion sensors for the computation of heading direction and time-to-contact. The application domain that we took into consideration for the implementation of such architectures, is the promising one of vehicle navigation. Having defined the types of images to be analyzed and the types of processing to perform, we were able to use a priori infor- VLSI Architectures for Computation of Heading Direction and Time-to-contact 721 mation to integrate selectively the sparse data obtained from the velocity sensors and determine the qualitative properties of the optical flow field of interest.

To exploit their properties at a system level, we developed parallel image processing architectures forapplications that rely mostly on the qualitative properties of the optical flow, rather than on the precise values of the velocity vectors. Specifically, we designed twoparallel architectures that employ arrays of elementary motion sensors for the computation of heading direction and time-to-contact. The application domain thatwe took into consideration for the implementation of such architectures, is the promising one of vehicle navigation. Having defined the types of images to be analyzed and the types of processing to perform, we were able to use a priori infor- VLSI Architectures for Computation of Heading Direction and Time-to-contact 721 mation to integrate selectively the sparse data obtained from the velocity sensors and determine the qualitative properties of the optical flow field of interest.

The value of scientific digital-image libraries seldom lies in the pixels of images. For the primary scientific analysis of these data, it is necessary to detect, measure, and classify every sky object. The learning algorithms are trained to classify the detected objects and can classify objects too faint for visual classification with an accuracy level exceeding 90 percent. This accuracy level increases the number of classified objects in the final catalog threefold relative to the best results from digitized photographic sky surveys to date.

The value of scientific digital-image libraries seldom lies in the pixels of images. For large collections of images, such as those resulting from astronomy sky surveys, the typical useful product is an online database cataloging entries of interest. We focus on the automation of the cataloging effort of a major sky survey and the availability of digital libraries in general. The SKICAT system automates the reduction and analysis of the three terabytes worth of images, expected to contain on the order of 2 billion sky objects. For the primary scientific analysis of these data, it is necessary to detect, measure, and classify every sky object. SKICAT integrates techniques for image processing, classification learning, database management, and visualization. The learning algorithms are trained to classify the detected objects and can classify objects too faint for visual classification with an accuracy level exceeding 90 percent. This accuracy level increases the number of classified objects in the final catalog threefold relative to the best results from digitized photographic sky surveys to date. Hence, learning algorithms played a powerful and enabling role and solved a difficult, scientifically significant problem, enabling the consistent, accurate classification and the ease of access and analysis of an otherwise unfathomable data set.