Faces represent complex, multidimensional, meaningful visual stimuli and developing a computational model for face recognition is difficult [43]. We present a hybrid neural network solution which compares favorably with other methods. The system combines local image sampling, a self-organizing map neural network, and a convolutional neural network. The self-organizing map provides a quantization of the image samples into a topological space where inputs that are nearby in the original space are also nearby in the output space, thereby providing dimensionality reduction and invariance to minor changes in the image sample, and the convolutional neural network provides for partial invariance to translation, rotation, scale, and deformation. The convolutional network extracts successively larger features in a hierarchical set of layers. We present results using the Karhunen-Loeve transform in place of the self-organizing map, and a multi-layer perceptron in place of the convolutional netwo...

With the current rapid growth in multimedia technology, there is an imminent need for efficient techniques to search and query large image databases. Because of their unique and peculiar needs, image databases cannot be treated in a similar fashion to other types of digital libraries. The contextual dependencies present in images and the complex nature of two-dimensional image data make the representation issues more difficult for image databases. An invariant representation of an image is still an open research issue. For these reasons, it is difficult to find a universal content-based retrieval technique. Current approaches based on shape, texture, and color for indexing image databases have met with limited success. Further, these techniques have not been adequately tested in the presence of noise and distortions. A given application domain offers stronger constraints for improving the retrieval performance. Fingerprint databases are characterized by their large size as well as nois...

Faces represent complex, multidimensional, meaningful visual stimuli and developing a computational model for face recognition is difficult (Turk and Pentland, 1991). We present a hybrid neural network solution which compares favorably with other methods. The system combines local image sampling, a self-organizing map neural network, and a convolutional neural network. The self-organizing map provides a quantization of the image samples into a topological space where inputs that are nearby in the original space are also nearby in the output space, thereby providing dimensionality reduction and invariance to minor changes in the image sample, and the convolutional neural network provides for partial invariance to translation, rotation, scale, and deformation. The convolutional network extracts successively larger features in a hierarchical set of layers. We present results using the Karhunen-Loève transform in place of the self-organizing map, and a multilayer perceptron in place of the convolutional network. The Karhunen-Loève transform performs almost as well (5.3 % error versus 3.8%). The multilayer perceptron performs very poorly (40 % error versus 3.8%). The method is capable of rapid classification, requires only fast, approximate normalization and preprocessing, and consistently exhibits better classification performance than the eigenfaces approach (Turk and Pentland, 1991) on the database

1 1. INTRODUCTION 1 2. INSTALLATION INSTRUCTIONS 4 2.1 INTALLATING FROM CD-ROM 4 2.2 HIERARCHICAL DIRECTORY STRUCTURE 5 2.2.1 SOURCE CODE SUBDIRECTORY 6 2.3 AUTOMATED COMPILATION UTILITY 7 2.4 SYSTEM INVOCATION 10 3. SOFTWARE DOCUMENTATION 12 3.1 DO HSF FORM; src/lib/hsf/form.c; do_hsf_form() 12 3.1.1 INITIALIZE FOR HSF FORM; src/lib/hsf/form.c; init_form() 12 3.1.1.1 READ FORM IMAGE; src/lib/image/readrast.c; ReadBinaryRaster() 12 3.1.1.2 READ FIELD TEMPLATE; src/lib/hsf/hsftmplt.c; read_hsftmplt() 18 3.1.2 PROCESS HSF FORM; src/lib/hsf/form.c; process_form2() 18 3.1.2.1 REGISTER FORM IMAGE; src/lib/hsf/reghsf.c; register_hsf2() 18 3.1.2.1.1 Read Reference Points; src/lib/mdg/readmfs.c;p readmfsint2() 19 3.1.2.1.2 Locate Hypothesized Points; src/lib/hsf/hsfpoint.c; hsfpoints() 20 3.1.2.1.3 Compute Distortion Parameters; src/lib/stats/lsq3.c; chknfindparam3() 22 3.1.2.1.4 Transform Form Image; src/lib/image/fitimage.c; f_fit_param3_image2() 24 3.1.2.2 REMOVE FORM; src/lib/hsf/rmform.c;...

A pictorial query specification technique that enables the formulation of complex pictorial queries for browsing through a collection of spatially-referenced images is presented. It is distinguished from most other methods by the fact that in these methods the query image specifies a target database image in its entirety whereas in our approach the query image specifies the combination of objects that the target database image should contain rather than being treated as a whole image. The query objects are represented by shape features although other features such as color, texture, or wavelets could also be used. Using our technique, it is possible to specify which particular objects should appear in the target images as well as how many occurrences of each object are required. Moreover, it is possible to specify the minimum required certainty of matching between query-image objects and database-image objects, as well as to impose spatial constraints that specify bounds on the distanc...

Abstract not found

Building upon the utility of connected components, NIST has designed a new character segmentor based on statistically modeling the style of a person's handwriting. Simple spatial features (the thickness of the pen stroke and the height of the handwriting) capture the characteristics of a particular writer's style of handprint, enabling the new method to maintain a traditional character-level segmentation philosophy without the integration of recognition or the use of oversegmentation and linguistic postprocessing. Estimates for stroke width and character height are used to compute aspect ratio and standard stroke count features that adapt to the writer's style at the field level. The new method has been developed with a predetermined set of fuzzy rules making the segmentor much less fragile and much more adaptive, and the new method successfully reconstructs fragmented characters as well as splits touching characters. The new segmentor was integrated into the NIST public domain Form-Ba...

A method for representing geographic symbols for storage and retrieval in an image database is presented. Symbols are characterized by a collection of features that describe their shape. Many of these geographic symbols are composed of a circle (or rectangle) enclosing one or more small shapes. A new representation of such symbols based on their interior with the shapes considered as holes, termed a negative symbol, is described. A set of shape features used to characterize geographic symbols is presented. These features are appropriate for symbols that are represented by their negation as well as for symbols that are composed of only one part and thus are not represented by their negation. Negative symbols along with these features have been used successfully to index maps in a map image database system. Results of experiments testing the accuracy of the database using these features are given. 1 Introduction Consider a database containing a large number of images each composed of se...

Financial institutions, insurance companies, and government agencies are all aggressively pursuing the integration of automated forms processing into their everyday work flows. To use existing optical character recognition (OCR) technology, the forms that are currently hand-keyed will probably need to be redesigned. This paper presents some of the quantitative results generated by a comprehensive study of three versions of a redesigned tax form. Analyses show that using separately spaced bounding character boxes to represent fields provides superior machine readability over fields without character boxes, fields containing vertical ticks (combs), and fields with adjoining character boxes. It is also shown that character boxes containing two vertically stacked ovals cause writers much more difficulty to complete than do empty character boxes. The analyses also provide quantitative proof that writer idiosyncratic responses on forms are the major source of errors within the recognition sy...

Abstract—In many biometric pattern-recognition problems, the number of training examples per class is limited, and consequently the sample group covariance matrices often used in parametric and nonparametric Bayesian classifiers are poorly estimated or singular. Thus, a considerable amount of effort has been devoted to the design of other covariance estimators, for use in limited-sample and high-dimensional classification problems. In this paper, a new covariance estimate, called the maximum entropy covariance selection (MECS) method, is proposed. It is based on combining covariance matrices under the principle of maximum uncertainty. In order to evaluate the MECS effectiveness in biometric problems, experiments on face, facial expression, and fingerprint classification were carried out and compared with popular covariance estimates, including the reguralized discriminant analysis and leave-one-out covariance for the parametric classifier, and the Van Ness and Toeplitz covariance estimates for the nonparametric classifier. The results show that, in image recognition applications whenever the sample group covariance matrices are poorly estimated or ill posed, the MECS method is faster and usually more accurate than the aforementioned approaches in both parametric and nonparametric Bayesian classifiers. Index Terms—Bayesian classifiers, biometric recognition, covariance estimate, limited sample sizes, maximum entropy. I.