Image analysis – Applications – Target tracking or detecting
Reexamination Certificate
1999-10-22
2003-09-02
Au, Amelia M. (Department: 2623)
Image analysis
Applications
Target tracking or detecting
C382S190000, C706S052000
Reexamination Certificate
active
06614917
ABSTRACT:
1. BACKGROUND OF THE INVENTION
1.1 Field of the Invention
The present invention pertains to identifying objects in three-dimensional imagery data and, more particularly, an automatic target recognition system employing a dynamic process to identify targets in LADAR data.
1.2 Description of the Prior Art
Automatic target recognition (“ATR”) systems identify objects represented in multi-dimensional data to determine whether they are potential targets. Multi-dimensional data may be acquired in numerous ways. Laser detection and ranging (“LADAR”) systems are commonly employed for this purpose. In such systems, a laser signal is transmitted from a platform and, upon encountering an object, is reflected back to the platform. The platform then processes the reflected signal to obtain multi-dimensional data regarding the object causing the reflection. This data captures a number of the object's features such as its height, length, width, average height, etc. Since the platform transmits many laser signals across a general area potentially containing a number of objects reflecting the signals, it is necessary for the ATR system to examine the data to determine which reflecting objects might be of interest.
ATR systems are often divided into four subsystems: detection, segmentation, feature extraction, and identification. Identification is the final process which takes inputs such as the aforementioned features and establishes an identity for the object based on comparisons to previously determined features of known objects. The accuracy of the identification depends on several factors including the accuracy of the object features used in the comparison and the number of known objects constituting potential identifications.
The ATR system would ideally be able to compare completely accurate feature measurements against those of all known objects to establish the unknown object's identity. However, identification is frequently hampered by poor measurements such that, even if one of the potential identifications is correct, a complete match cannot be recognized. For instance, a length measurement might be compromised if part of the unknown object is obscured by a wall or fence and a width measurement might be affected by the orientation of the object relative to the platform. ATR system design constraints for size and speed may also affect performance.
An ATR system must therefore, as a practical matter, quickly establish the best possible identity with available computing resources. However, conventional computing techniques are poorly suited to meet these design constraints. For instance, conventional computing resources classically analyze information in two states such as “yes” and “no”, or “true” and “false” An incomplete match in the comparison phase of a suitable operational ATR system cannot easily be represented in two states and consequently requires extensive analysis to produce a useable answer. The extensive analysis, in turn, consumes time and computing resources. This consumption becomes more acute as the list of potential identifications becomes more comprehensive since there typically will be no or, at most, only a few complete matches. Conventional computational techniques are consequently poorly adapted to meet the various design constraints in practically implementing an ATR system and, as demand for improved system performance rises, become still less suited. Thus, there is a need for a new method for identifying objects in three-dimensional data which can dynamically account for conditions prevailing when the multi-dimensional data is collected.
One novel approach is presented in commonly owned U.S. Pat. No. 5,893,085, entitled “Dynamic Fuzzy Logic Process for Identifying Objects in Three Dimensional Data” Disclosed therein is a method of weighting the importance of particular feature comparisons and shifting particular feature membership functions based on the level of confidence in the accuracy of the feature measurement. This method is certainly a major improvement over prior art, but leaves room for improvement in the area of the adjustments made to the feature comparison process in view of not only confidence in the feature measurement itself, but in view of the conditions and circumstances prevailing at the time of the data collection which affect the expected range of values a known feature and the feature value should assume under similar conditions and circumstances.
2. SUMMARY OF THE INVENTION
The present invention in one embodiment is a method for identifying an object in observation-related data of an object in a surrounding environment. The method includes (a) providing (1) a set of comparative feature values associated with one or more known objects in one or more specified observation conditions and (2) a set of observation-related values including (i) a set of feature values for one or more features of an object as observed in an environment, (ii) a set of condition values for one or more conditions of the observation, including one or more of an orientation value, a depression value, a resolution value, and an obscuration value; and (b) generating one or more identification values for the object by comparing the set of feature values to the set of comparative feature values, including compensating for the set of condition values.
REFERENCES:
patent: 5640468 (1997-06-01), Hsu
patent: 5893085 (1999-04-01), Phillips et al.
patent: 0884603 (1998-12-01), None
Au Amelia M.
Lockheed Martin Corporation
Miller Martin
LandOfFree
Dynamic process for identifying objects in multi-dimensional... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Dynamic process for identifying objects in multi-dimensional..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Dynamic process for identifying objects in multi-dimensional... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3022706