Image analysis – Applications
Reexamination Certificate
1999-08-03
2001-10-16
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
Reexamination Certificate
active
06304664
ABSTRACT:
FIELD OF INVENTION
The present invention relates to multispectral imaging. In particular, the present invention teaches a variety of systems and methods for facilitating the separation of ocean surface reflected light, light scattered in the ocean and atmosphere, and anomalous objects in multispectral ocean imagery. The intent of the processing is to detect the anomalous objects, which may be either on or below the ocean surface.
BACKGROUND OF THE INVENTION
Detection of low contrast objects in the ocean has many applications including environmental monitoring, locating schools of fish and sunken ships, and aiding search and rescue operations. Furthermore, military applications include countermine measures (CMM) and anti-submarine warfare (ASW). In addition, navigation applications include the detection and avoidance of navigation hazards, and the interpretation of ocean bottom topographies, such as reefs.
Multispectral images taken from above the ocean's surface generally contain background clutter consisting of light reflected from the ocean surface, light scattered in the atmosphere above the ocean, and upwelling from scattering the water column. Relative to the background clutter, the anomalous objects of interest are generally of very low contrast. The challenge is in removing the pervasive background light so as to render anomalous surface or underwater objects more visible.
Most ocean clutter is due to light reflected from the surface. The prior art to filter ocean clutter includes two methods; one based on temporal averaging of multiple monochromatic images, and the second using multispectral imaging. Temporal averaging uses a video camera. The sequence of images is corrected for frame-to-frame perspective, translational, rotational and magnification changes before stacking (integration). The integration approach works because surface light is modulated at the frequency of the surface waves. Integrating over a time period spanning the wave period, or longer, decreases the surface light. Typical ocean waves of order of 100 m can be filtered with 10-second or longer integration. Shorter waves, order of 1 m, can be filtered just as well in as little as 3 seconds. The method is effective in removing only the surface reflection. It does not eliminate the upwelling light clutter.
The prior art multispectral method makes a global estimate of the background light spectrum and subtracts the estimate pixel-by-pixel. This method implicitly assumes that the background spectrum can be described by a global 1-component spectrum. Prior art as depicted in
FIGS. 1A and 1B
illustrats multispectral imaging system configurations. The systems use data collected by multispectral imaging systems flown over the ocean, on satellites, (
FIG. 1A
) or aircraft (FIG.
1
B). They work in the daytime, using sunlight to illuminate the object of interest. The imaging camera is usually staring down (nadir view). Typically, the camera is a high quality CCD imaging camera that simultaneously images in several (typically 2 to 10) spectral bands (multispectral), or up to hundreds of spectral bands (hyperspectral).
Thus, there have generally been two different methods utilized for removing the unwanted light reflection in an ocean image. One exploits multiple spectral bands, and the other uses time integration. The choice of which is used leads to different sensor designs—one is a multispectral or hyperspectral imager; the other is simply a long exposure or video camera (possibly with a select narrow spectral band filter).
However, a 1-componenet model does not accurately capture the spectral variability of the clutter because the light in each image pixel is actually a mixture of several components, each having a different spectrum, in relative amounts that can vary from one pixel to another. It is highly desirable to have a process that represents the reflected light more accurately than is possible utilizing a 1-component process. The desired process would subtract much more light clutter, thus making fainter objects easier to detect. In addition, the desired process should allow faster processing of the multispectral image data.
SUMMARY OF INVENTION
The present invention addresses the aforementioned characteristics by providing a de-glinting process that utilizes a two component model to estimate reflected light from the ocean surface and scattered light from the water column below the surface. The advantage of the present invention over the prior art 1-component model is that the 2-component model of the present invention more accurately represents the reflected light. These enables the present invention to subtract much more light clutter than is possible utilizing a 1-component model, making fainter objects easier to detect. In addition, the present invention is generally able to process image data faster than previous 1-component processes. Under certain conditions, the present invention is able to process image data 100 times faster than 1-component processes.
The system includes an input device for obtaining the image data, a device, such as a computer, for analyzing the image data and applying the two-component process, and an output device for outputting the clutter subtracted image.
REFERENCES:
patent: 5243541 (1993-09-01), Ulich
patent: 5371542 (1994-12-01), Pauli et al.
patent: 5661817 (1997-08-01), Hatlestad et al.
patent: 5805106 (1998-09-01), Baum
patent: 6028672 (2000-02-01), Geng
Abileah Ronald
Silva Dennis M.
Johns Andrew W.
Nakhjavan Shervin
Oppenheimer Wolff & Donnelly LLP
SRI - International
LandOfFree
System and method for multispectral image processing of... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for multispectral image processing of..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for multispectral image processing of... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2555842