Image analysis – Pattern recognition – Classification
Reexamination Certificate
1997-09-26
2001-03-06
Mehta, Bhavesh (Department: 2721)
Image analysis
Pattern recognition
Classification
C382S181000
Reexamination Certificate
active
06198847
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a pattern recognition apparatus, and more particularly, to a pattern recognition apparatus for controlling parameters in order to efficiently recognize a pattern when an input device included in the apparatus undergoes various kinds of restrictions which can be changed by the parameters.
2. Description of the Related Art
In an image recognition apparatus provided in a robot moving in a three-dimensional space, since it is unknown in which direction an image signal to be recognized is present, the apparatus is requested to input a signal within a range as wide as possible. At the same time, the apparatus is required to have sufficient spatial resolution to recognize a certain pattern. As an image input device which simultaneously satisfies such requirements, a device, which performs nonuniform sampling such that, referring to the characteristics of a human retina, an image near the center of an optical axis is sensed with high resolution and resolution decreases as an image is separated from the optical axis, has been devised.
In such nonuniform sampling, in order to exactly recognize a pattern sampled with low resolution at a portion surrounding an input image, it is necessary to change the optical axis so that the pattern is again sampled in a high-resolution region near the center of the optical axis. That is, nonuniform sampling becomes an effective input method only with optical-axis control.
A method for controlling an optical axis based on features of an input image input according to nonuniform sampling has been devised as an optical-axis control method for the above-described purpose. For example, the absolute value of the slope of an image intensity, an output value after passing through a two-dimensional filter, and the like are used as the features. An optical-axis control method based on a knowledge base has also been devised. In this knowledge-base method, tasks to be executed by a system, and a set of images or patterns to be recognized are expressed by a Bayes network, and an operation to be subsequently performed is determined based on the probability structure in order to efficiently execute given tasks.
As described above, most methods of conventional optical-axis control are based on image signals. Accordingly, an optical axis is controlled only by geometric properties of an image, such as a portion having a large edge intensity or a portion having a large filter output. In such control methods, an optical axis is moved to a portion which is not required for a task if the absolute value of the slope is large. Furthermore, in the above-described knowledge-base method, since the Bayes network is used, the structure of data which can be expressed is limited.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a pattern recognition method and apparatus which can perform optical-axis control without depending only on geometric properties of an image.
It is another object of the present invention to provide a pattern recognition method and apparatus which can more exactly recognize a pattern.
It is still another object of the present invention to provide a pattern recognition method and apparatus which can correct an optical axis at a high speed.
According to one aspect, the present invention which achieves these objectives relates to a pattern recognition apparatus comprising input means for inputting a first signal representing a pattern to be recognized, sampling means for nonuniformly sampling the first signal input from said input means, recognition means for recognizing the pattern represented by the first signal based on a second signal obtained by the sampling by the sampling means, and changing means for changing a sampling position in nonuniform sampling by the sampling means based on a result of the recognition by the recognition means.
According to another aspect, the present invention which achieves these objectives relates to a pattern recognition method comprising the steps of inputting a first signal representing a pattern to be recognized, nonuniformly sampling the input first signal, recognizing the pattern represented by the first signal based on a second signal obtained by the sampling, and changing a sampling position in nonuniform sampling in the sampling step based on a result of the recognition.
According to still another aspect, the present invention which achieves these objectives relates to a storage medium storing a pattern recognition program comprising the steps of inputting a first signal representing a pattern to be recognized, nonuniformly sampling the input first signal, recognizing the pattern represented by the first signal based on a second signal obtained by the sampling, and changing a sampling position in nonuniform sampling in the sampling step based on a result of the recognition.
Other objectives and advantages besides those discussed above shall be apparent to those skilled in the art from the description of the preferred embodiments of the invention which follows. In the description, reference is made to accompanying drawings, which form a part thereof, and which illustrate examples of the invention. Such examples, however, are not exhaustive of the various embodiments of the invention, and therefore reference is made to the claims which follow the description for determining the scope of the invention.
REFERENCES:
patent: 5555102 (1996-09-01), Dalton
patent: 5649021 (1997-07-01), Matey et al.
patent: 5835641 (1998-11-01), Sotoda et al.
“Centering Peripheral Features In an Indoor Environment Using a Binocular Log-Polar 4 DOF Camera Head”, R.A. Peters, et al., from International Workshop on Biorobotics: Human-Robot Symbiosis, Tsukuba, Japan, May 1995, vol. 18, No. 1-2, published in,Robotics and Autonomous Systems 18, Jul. 1996, Elsevier, Netherlands, pp. 271-281.
“Active Object Recognition Integrating Attention and Viewpoint Control”, S.J. Dickson, et al., Computer Vision—ECCV '94. Third European Conference on Computer Vision, vol. II, pp. 3-14, 1994, Berlin, Germany.
Canon Kabushiki Kaisha
Fitzpatrick ,Cella, Harper & Scinto
Mehta Bhavesh
LandOfFree
Apparatus and method for recognizing a nonuniformly sampled... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for recognizing a nonuniformly sampled..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for recognizing a nonuniformly sampled... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2547231