Showing papers on "Sketch recognition published in 1992"
01 Jan 1992
TL;DR: Three viewpointdependent aspects of human performance in recognition are surveyed: canonical views, mental rotation, and limited anisotropic generalization to novel views by analyzing the functioning of an implemented network model of recognition.
Abstract: Unlike basic-level categorization, which is largely viewpoint-invariant, object recognition at the subordinate levels depends on the observer's point of view in several ways. The first part of this article surveys three viewpointdependent aspects of human performance in recognition: canonical views, mental rotation, and limited anisotropic generalization to novel views. The second part offers a detailed but informal computational account of these phenomena, obtained by analyzing the functioning of an implemented network model of recognition. The success of the model in replicating central features of human performance supports the notion that at least one of the available pathways to recognition in the human visual system relies on viewpoint-specific representations.
8 citations
30 Aug 1992
TL;DR: This paper presents work on the extraction of temporal information from static images of handwriting and its implications for character recognition.
Abstract: Handwritten character recognition is typically classified as online or offline depending on the nature of the input data. Online data consists of a temporal sequence of instrument positions while offline data is in the form of a 2D image of the writing sample. Online recognition techniques have been relatively successful but have the disadvantage of requiring the data to be gathered during the writing process. This paper presents work on the extraction of temporal information from static images of handwriting and its implications for character recognition. >
8 citations
07 Jun 1992
TL;DR: The authors survey a successful replication of central characteristics of performance in 3-D object recognition by a computational model based on interpolation among a number of stored views of each object.
Abstract: The topics discussed here are network models of object recognition; a computational theory of recognition; psychophysical support for a view-interpolation model: and an open issue, features of recognition. The authors survey a successful replication of central characteristics of performance in 3-D object recognition by a computational model based on interpolation among a number of stored views of each object. Network models of 3-D object recognition based on interpolation among specific stored views behave in several respects similarly to human observers in a number of recognition tasks. Even closer replication of human performance in recognition should be expected, once the issue of the features used to represent object views is resolved. >
8 citations
30 Aug 1992
TL;DR: A transputer-based parallel machine for handwritten character recognition is proposed and an algorithm based on structural features and on a tree classifier was used to accomplish the pre-classification of the unknown sample in order to speed up the recognition process.
Abstract: A transputer-based parallel machine for handwritten character recognition is proposed. An algorithm based on structural features and on a tree classifier was used to accomplish the pre-classification of the unknown sample in order to speed up the recognition process. The algorithm for the final classification is based on the description of the strokes through Fourier descriptors. The learning phase is accomplished through a man-machine interactive process. The proposed system can expand its knowledge base. A special representation of this knowledge base is proposed in order to record a great amount of data in a suitable way. A fast multistroke handwritten isolated character recognition system is presented. The test of this system was performed on a PC based prototype while the realization of a parallel transputer based working machine is in progress. Experimental results obtained applying these machines to handwritten numerals recognition are reported. >
7 citations
07 Jun 1992
TL;DR: A pattern recognition system that autonomously learns to categorize and recognize patterns independently of their position in an input image by combining higher-order with first-order networks and the mechanisms known from ART.
Abstract: A proposal by M. B. Reid et al. (1989) to improve the efficiency of higher-order neural networks was built into a pattern recognition system that autonomously learns to categorize and recognize patterns independently of their position in an input image. It does this by combining higher-order with first-order networks and the mechanisms known from ART. Its recognition is based on a 16*16 pixel input which contains a section of the image found by a separate centering mechanism. With this system position invariant recognition can be implemented efficiently, while combining all the advantages of the subsystems. >
5 citations
07 Jun 1992
TL;DR: The author discusses two complete neural network recognition systems, a character recognition system and a fingerprint classification system that demonstrated state-of-the-art accuracy but both need improvements to be commercially viable.
Abstract: The author discusses two complete neural network recognition systems, a character recognition system and a fingerprint classification system. The requirements for a total vision system must include the capability for image isolation, segmentation, and feature extraction, as well as recognition. The systems were developed on a massively parallel array processor which was used to illustrate the importance of these higher-level functions. Both of these systems demonstrated state-of-the-art accuracy but both need improvements to be commercially viable. The issue in the character recognition system is to provide this accuracy at a speed compatible with commercial requirements of 1 page/s. This will require more sophisticated higher-level image parsing functions without loss of accuracy. The issue in fingerprint classification is the requirement for 99.7% accuracy at current speeds. >
5 citations
30 Aug 1992
TL;DR: The application of the structural learning technique known as error correcting grammatical inference to planar shape recognition is discussed and illustrated with a non-trivial printed digit recognition task.
Abstract: The application of the structural learning technique known as error correcting grammatical inference to planar shape recognition is discussed and illustrated with a non-trivial printed digit recognition task. Experimental results are presented and compared with those of other more conventional (non-structural) techniques, showing the new technique to provide significantly improved performance. >
4 citations
30 Aug 1992
TL;DR: The authors propose an approach to define objects qualitatively and hierarchically by generic shapes (primitives) arranged by generic relations so that a class of objects has the same description and to recognize them in a parallel and bottom-up way in the image.
Abstract: The authors propose an approach to define objects qualitatively and hierarchically by generic shapes (primitives) arranged by generic relations so that a class of objects has the same description and to recognize them in a parallel and bottom-up way in the image. >
3 citations
30 Aug 1992
TL;DR: An original algorithm developed for specific applications where performing omnifont text recognition is needed but where only limited computer resources are available in terms of memory and CPU power is discussed.
Abstract: The problem of optical character recognition (OCR) has been studied since the early stages of the computer industry. However, the problem of omnifont text recognition (OTR) (recognition of any kind of document without having information on the typeface) is quite recent. Omnifont text recognition is usually CPU intensive and requires a large amount of memory. This paper discusses an original algorithm developed for specific applications where performing omnifont text recognition is needed but where only limited computer resources are available in term of memory and CPU power. >
2 citations
TL;DR: A new architecture is presented for multi-map, self-organizing pattern recognition which allows concurrent massively parallel learning of features using different maps for each feature type similar to the multi- map structures known to exist in the vertebrate sensory cortex.
Abstract: A new architecture is presented for multi-map, self-organizing pattern recognition which allows concurrent massively parallel learning of features using different maps for each feature type The method used is similar to the multi-map structures known to exist in the vertebrate sensory cortex The learning used to update memory locations uses a feed-forward mechanism and is self-organizing The architecture is described by the acronym FAUST (Feed-forward Association Using Symmetrical Triggering) As a demonstration of the effectiveness of FAUST, a character recognition, fingerprint classification, and forms recognition programs have been constructed on a massively parallel compute The character recognition program can perform 99% accurate character recognition on medium-quality machine printed digits at a speed of 24 ms/digit, and 88% recognition on multiple-writer handprint with a 23% substitutional error rate The form recognition program can achieve 94% accuracy on complex forms The fingerprint classification program classifies 93% of fingerprints correctly with 10% rejection rate All of the calculations were performed on an Active Memory Technology DAP 510
30 Aug 1992
TL;DR: The author presents an approach to automatic recognition of design concepts, and the method of relaxation recognition is introduced to complete the recognition of abstract patterns.
Abstract: Pattern recognition studies have been traditionally concerned with concrete patterns such as two-dimensional images or graphics and three-dimensional objects. Little attention has been paid to abstract patterns such as design concepts and mathematical arguments. This paper attempts to address the abstract pattern recognition problem. The author presents an approach to automatic recognition of design concepts. Entities and relations are proposed for the description of design concepts. Computer algorithms are developed to capture the concepts in the design. The method of relaxation recognition is introduced to complete the recognition of abstract patterns. >
18 Oct 1992
TL;DR: Object recognition is considered by considering it in the context of an agent performing it in an environment, where the agent's intentions translate into a set of behaviors, and relative depth was used to recognize the different stages in the process.
Abstract: The authors propose to study object recognition by considering it in the context of an agent performing it in an environment, where the agent's intentions translate into a set of behaviors. The problem becomes a problem of action from intensity functions. In accomplishing a behavior, the next step of action from the images is determined. Acquiring the information for action is a solution for a recognition task. The recognition task is agent and behavior dependent and can use the output of different visual modules. The implementation of one visual module and its use for purposive recognition are described. It is shown how to robustly extract relative depth from a stereo setup without correspondence and calibration, and how this visual module can be used under some intentions and behaviors. For recognition under navigation behavior, relative depth was used to recognize obstacles by isolating unexpected objects in close range. For grasping behavior relative depth was used to recognize the different stages in the process. >
30 Aug 1992
TL;DR: The main aim of the system is to perform an object recognition based on contour description and being independent of noise and indeterminations in the digitization process.
Abstract: The interest in fuzzy algorithms is increasing in a wide range of pattern recognition applications. This paper describes a recognition system using fuzzy algorithms and parameters, with particular enhancement in the system architecture. This system is able to recognize flat polygonal objects in real time. The main aim of the system is to perform an object recognition based on contour description and being independent of noise and indeterminations in the digitization process. This system is made up of three modules. The first one is a break points detector. The second block performs a contour description by means of fuzzy parameters that allows us to recognize the picture in the third block. >
10 Nov 1992
TL;DR: An approach is presented to the invariant recognition of objects under dynamic perceptual conditions by the closed-loop integration of recognition processes of computer vision together with an incremental machine learning process.
Abstract: An approach is presented to the invariant recognition of objects under dynamic perceptual conditions. In this approach, images of a sequence are used to adapt object descriptions to perceived online variabilities of object characteristics. This adaptation is made possible by the closed-loop integration of recognition processes of computer vision together with an incremental machine learning process. The experiments presented were run for the texture recognition problem and were limited to a partially supervised evolution of concept descriptions (models) rather than utilizing a fully autonomous model evolution. Obtained results are evaluated using the criteria of system recognition effectiveness and recognition stability. >