Image analysis – Applications – Target tracking or detecting
Reexamination Certificate
2001-07-27
2004-06-22
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
Target tracking or detecting
C382S291000, C348S143000
Reexamination Certificate
active
06754368
ABSTRACT:
FIELD OF INVENTION
This invention relates to an object recognition and tracking system.
BACKGROUND
Systems have been proposed for counting objects such as people passing along a corridor in for instance a shopping centre but such systems are not adapted for recognition or discrimination of certain or different types of objects and their position and to track such objects over time.
It is the object of this invention to provide an object recognition and tracking system.
BRIEF DESCRIPTION OF THE INVENTION
In one form therefore, although this may not necessarily be the only or broadest form invention is said to reside as an object recognition and tracking system including at least one detector array, the or each detector array including a first type of detector and a second type of detector, the first type of detector being adapted to detect the presence of an object in a detector space and to define the object's position, the second type of detector being adapted to use the defined object position to identify or recognise the object or to count or to record the presence of the object if it is of a selected type.
The presence of an object as determined by the first detector may be detected by a change in contrast boundaries or by a change in pattern or by movement and hence to detect objects that move within the detector space.
The detectors may be adapted to use a portion of the electromagnetic spectrum selected from radar, microwave, radio frequency, infrared, microwave, millimetre wave, light or other portion of the electromagnetic spectrum or to use sonar or some other detection system.
The first type of detector and the second type of detector may be of the same type or they may be of different types. Hence the first type of detector may be a low resolution video system and the second type of detector may be a radio frequency detection system.
Where the first detector and the second detector are in effect the same detector the second detector may operate at a higher resolution than the first detector or be a specific identifying type.
In a preferred form of the invention the first type of detector maybe of an insect eye type. That is, a detector which is adapted to detect change in contrast between adjacent pixels in a display space rather than to identify objects absolutely. In such a form of the invention the second type of detector may be a digital signal processor detector such as a digital video detector. Hence the first type of detector acts to identify a specific region of interest within a detector space and then the digital signal processor detector need only identify the object by looking in that particular region of the space rather than by processing a whole video image of the detector space. By this means considerably less processing power is required and greater efficiencies can be obtained.
The first detector and the second detector could be the one device with the second detector acting as a second stage of the process.
Hence, in an alternate form of the invention the first and second types of detectors may be provided by a single video detector the observed image of which is handled in two different ways. In a first way the observed image may be digitised and stored in successive memory locations and then sequential images may be scanned to detect changes in the digitised image over successive frames which would indicate a moving contrast boundary. This thereby defines a moving object in the detector space and the same digital image can then be analysed in that particular region of the detector space by the second type of detector. The second type of detector could use the subsequent or previous frames to improve the recognition or result with respect to the object of interest.
If the recognition process does not reach predetermined or artificially learned threshold levels of reliability then further stored or downloaded algorithms could be used in an iterative process until the results are improved.
Whilst the first detector type is used to identify an object of interest, the second type of detector is used for recognition purposes. Each object possesses a number of characteristics such as mass, height, shape, contrast, brightness, colour, patterns, velocity, heat, reflectivity, signature and many others. Each characteristic on its own does not necessary lend itself as being a characteristic that can be used to recognise an object by itself, and some characteristics lend themselves to be more easily identified than other characteristics. A preferred part of this invention is the analysis of the most recognisable of characteristics first, then by using other characteristics and successive frames in time, at each step in the iterative analysis process, the system improves its likelihood of recognising an object with greater accuracy. The determination of which characteristics are analysed first and the sequence of which they are analysed is a most preferred part of this invention.
In an alternate form of the invention, the system may be built with artificial intelligence to make these decisions without external intervention. In this case, the system determines which characteristic would be analysed first. As time progresses, the system may identify new characteristics or a new sequence of characteristics that may better identify an object of interest and therefore become self-improving. In this case, self improved procedures, or algorithms are stored with an existing memory resident library for future reference. The system would use artificial intelligence analysis to decide which algorithm is more suited by its success and speed of recognition using probability analysis.
Generally, therefore, the invention discloses a system which may be broken down into a number of stages. The first stage of the process would include an object detection engine to capture and load the digitised image sequences from the hardware into an appropriate digital format for the object recognition engine. The first type of detector may identify the boundary of an object as determined over successive frames as the object moves across the detector space. The second type of detector, which could form a second stage of the system would be an object recognition engine used to, firstly, convert certain characteristics into useful mathematical images or algorithms, and secondly, may be used to perform and record results over time necessary to analyse specific characteristics of the objects in question. The characteristics under analysis may include but would not be limited to one or more of items such as the mass, height, shape, colour, pattern, contrast, texture, heat, reflectivity, frequency, velocity, signal radiation, signature etc. Each characteristic of the detected object is given a chance of probability of being an object of interest. To obtain higher levels of accuracy, successive digitised images may be analysed over time using stored procedures, stored libraries, analysis algorithms and an iterative analysis process which can manage the observation of multiple characteristics from multiple objects in real-time.
Each analysis iteration increases (or decreases as the case may be) the chance of a detected object being an object of interest. If the level of probability reaches a predetermined level of acceptance or threshold level, the object is recognised as an object of interest. To improve the probability analysis, additional kinds of the second type of detector could be included within a single detector space used to view different characteristics of an object of interest to improve the probability of recognising an object.
Additional detectors types may be positioned in different locations and facing a different angles but mounted to view the same detector space to improve the chance of differentiating objects, placing or positioning objects in a three dimensional space or locating an object which would otherwise be obscured by the use of only one detector type.
The second type of detector may be adapted to interrogate an interrogatable device associated with the object. I
Beonic Corporation Pty. Ltd
Johns Andrew W.
LandOfFree
Object recognition and tracking system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Object recognition and tracking system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Object recognition and tracking system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3339135