Computationally enhanced image recognition system
AssigneeThe United States of America as represented by the Secretary of the Air Force
InventorsDaniel W. Repperger, Alan R. Pinkus
Life of this patent
Estimating the term…
Abstract
Method for enhancing visual images. An operator manipulates a visual image by adjusting darkness levels within each primary color to embellish the recognition of distinct objects which may appear in the visual rendering. The overall system can be implemented as a ground-based system where the operator can adjust the algorithm to manipulate the darkness levels of a certain number, N, of pixels. A second implementation can be achieved in a helmet mounted display or glasses donned by an operator.
Claims15
An image enhancement method comprising the steps of:
- receiving visual data of a desired object within a compressed image;
- transmitting said visual data to a central processing unit;
- decomposing said visual data into histograms of frequency of pixels versus darkness level within each primary color red, blue and green;
- selecting the most highly uncertain color from said decomposing step;
- constructing a parabolic template with the vertex of the parabola appearing at the mean distribution of the total area under the histogram curve;
- overlaying a true histogram from said visual data and said parabolic template;
- generating a spatial gradient of all candidate pixels;
- moving N number of pixels with maximum gradients through operator adjustment based on visual enhancement;
- and
- transferring an output signal of an enhanced visual image of said desired object from said central processing unit to said operator.
An image enhancement method comprising the steps of:
- receiving visual data of a desired object within a compressed image using multi-spectral sensors from an operator wearing a head-mounted display;
- transmitting said visual data to a central processing unit;
- decomposing said visual data into histograms of frequency pixels versus darkness level within each primary color red, blue and green and selecting the most highly uncertain color;
- constructing a parabolic template with the vertex of the parabola appearing at the mean distribution of the total area under the histogram curve;
- overlaying a true histogram from said visual data and said parabolic template comprising the step of:
- selecting a threshold of 0.5 for a darkness level varying between 0 and 1;
- first setting a darkness level to zero when, for each pixel, the darkness level is less than or equal to said threshold;
- and
- second setting a darkness level to one, when for each pixel, the darkness level is greater than the threshold;
- generating a spatial gradient of all candidate pixels;
- moving N number of pixels with maximum gradients through operator adjustment knob manipulation;
- and
- transferring an output signal of an enhanced visual image of said desired object from said central processing unit to said operator.
Dependencies are inferred from the claim text.