Radiant energy – Photocells; circuits and apparatus – Photocell controlled circuit
Reexamination Certificate
1999-06-14
2002-03-12
Le, Que T. (Department: 2878)
Radiant energy
Photocells; circuits and apparatus
Photocell controlled circuit
C250S2140RC
Reexamination Certificate
active
06355922
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to imaging systems. More specifically, the present invention relates to systems and methods for finding detection thresholds in infrared imaging systems that separate real scene objects observed through an optical system from sensor fixed artifacts.
2. Description of the Related Art
Infrared image sensors are used in military, astronomical and other applications. For example, in military applications, infrared image sensors are often used for target acquisition and tracking. The seekers of missiles often use arrays of image detectors sensitive to energy in the visible or (more typically) infrared portion of the electromagnetic spectrum. Unfortunately, these arrays, referred to as focal plane arrays are subject to anomalies such as detector to detector nonuniformity in sensitivity and gain and fixed pattern noise. While numerous techniques are known in the art for addressing detector to detector nonuniformity in sensitivity and gain, fixed pattern noise continues to be problematic.
Fixed pattern noise (FPNs) are sensor fixed artifacts induced by non-uniform response of the focal plane array (FPA). The non-uniform response causes the FPA output to be spatially varying even when illuminated by a uniform source.
That is, when viewed through an infrared imager, many objects (for example, airplanes, vehicles, building, etc) consist of several regions which are locally homogeneous in intensity. The process of finding those pixels in an image which correspond to an object is called ‘segmentation’. For purposes of guiding a vehicle to an object, one needs to locate the position of the object in the image and provide this position to a guidance and control system at a specified (high) data rate and within a predefined time interval (latency requirement). Segmentation of the object is performed on the image to define the object. The centroid of the segmented object is then reported as the position of the object.
Frequently, one or more of the regions on an object has high contrast and can be easily detected (segmented). The centroid of these detected regions can be used for guidance purposes. As stated earlier, these positions must be reported at high data rate. For the purposes of guidance, it is sufficient to guide to a high contrast region on the object. It is often desirable to find the centroid of the entire object (or as much of it as possible) as the guidance point. To accomplish this, it is necessary to segment the entire object. As long as the high contrast regions are being found at a sufficiently high data rate to satisfy guidance, there is no data rate requirement on segmenting the entire object because it is not used for guidance. Once a larger portion of the object has been segmented, the guidance system is told to shift its guidance point on the centroid of the entire object. Thus, the process of segmenting out the entire object may take many image frames.
The detection of the high contrast regions may be easily accomplished with a high detection threshold. With a high detection threshold, there is little likelihood of segmenting out pixels that belong to the background and linking them to the object of interest. To detect the lower contrast regions of an object, it is necessary to lower the detection threshold. When the detection threshold is lowered, it is possible that background pixels adjacent to the object may also exceed the detection threshold. Because all threshold pixels that are spatially touching (either 4 connected or 8 connected as known to those skilled in the art) are considered as one object, too low a threshold then causes the object to merge with the background and may lead to a centroid that is not on the object of interest. Providing the centroid of the merged object/background to guidance will then cause a miss.
The question to be addressed is this, how can the threshold be set to be low enough to detect the lower contrast regions of the object of interest without simultaneously setting the threshold so low that it merges adjacent background in with the object. Of course, the contrast of portions of the object may be so low that it is not possible to segment all of the regions without merger. In this case, the objective is to segment as much of the object as possible without segmenting adjacent background pixels.
The conventional technique for object segmentation is based on finding a threshold that leads to a stable area. That is, the threshold is changed until a derivative of the area segmented with respect to threshold is essentially zero. The principle of this method assumes that that there is a significant difference in grey level between the object of interest (the target) and its adjacent regions in the image. Hence, once the threshold reaches the correct value, further decreases in the threshold (assuming the target is brighter than its neighbors) does not lead to further increases in area.
Unfortunately, the conventional technique for object segmentation fails to include all of the target when it is made up of multiple locally homogeneous regions because it will segment only one local region. In addition, when adjacent regions have similar grey levels as the target due to image artifacts and has been included in the segmentation, there is no method of rejection.
Hence, a need remains in the art for a system and method for object segmentation for example, in infrared image when sensor fixed artifacts are present. More specifically, there is a need in the art for a system and method for segmenting as much of the target as possible without merging other objects in with the target through setting of a threshold that is too low.
SUMMARY OF THE INVENTION
The need in the art is addressed by the present invention which provides A system and method of finding a detection threshold that separates real scene objects seen through an optical system from sensor fixed artifacts. The inventive system continuously scans the sensor so that objects in the scene move in an inverse sense relative to the scanning motion while fixed frame artifacts remain fixed in focal plane coordinate system. The difference in temporal behavior is used to discriminate against fixed frame artifacts. The inventive system finds a detection threshold that maximizes the portion of a target that may be segmented while avoiding setting the threshold so low that it merges spatially adjacent artifacts in with the target.
In the illustrative embodiment, the inventive system includes a scanning sensor for providing a video signal comprising plural image frames representative of a scene having a rigid body therein. The system uses first circuit to detect a first region of high contrast in the scene using a high threshold and calculates a first centroid in response thereto. A second circuit is used to detect a second region using a low threshold and to calculate a second centroid in response thereto. The centroids are compared and a difference vector is calculated therebetween. The difference vector is analyzed over plural image frames and to segment a larger region of the rigid body than that afforded using conventional teachings. If the difference vector remains substantially unchanged over a predetermined number of image frames a lower threshold is set. The process is performed periodically to grow the segmented pixels to the object in its entirety.
REFERENCES:
patent: 4972260 (1990-11-01), Fujikawa et al.
Jensen Stephen M.
Lo Thomas K.
Le Que T.
Lenzen, Jr. Glenn H.
Raufer Colin M.
Raytheon Company
Rudd Andrew J.
LandOfFree
Imaging system and method for hardbody detection and sensor... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Imaging system and method for hardbody detection and sensor..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Imaging system and method for hardbody detection and sensor... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2822934