Morphological automatic Landolt C orientation detection
Inventors
Pinkus, Alan R • Dommett, David W • Task, Harry Lee
Assignees
United States Department of the Air Force
Publication Number
US-9773183-B2
Publication Date
2017-09-26
Expiration Date
2036-05-05
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
A method and apparatus are provided for automatically assessing a resolution of an optical sensor. An image is obtained from the sensor. The obtained image is thresholded to generate a binary image of black and white pixels using a first threshold value. A target area is selected from the binary image. The first threshold value is increased to generate a second threshold value. The target is thresholded using the second threshold value. At least one pixel is tagged at a center of the target area. Pixels surrounding a tagged pixel are processed, tagging white pixels. If a tagged pixel is at an edge of the target area, an orientation of an object in the target area is determined based on that edge. And, if no pixel is tagged at the edge of the target area, the second threshold value is decreased and the thresholding, tagging, and processing steps are repeated.
Core Innovation
The invention provides a morphological approach for automatic detection of the orientation of Landolt C targets within digital images to assess optical sensor resolution. The method involves obtaining an image from the sensor, thresholding the image to create a binary image, selecting a target area, adjusting threshold values iteratively, tagging pixels at the target center, and processing surrounding pixels to determine the orientation based on edge conditions.
This approach is more computationally efficient and accurate than existing methods and enables automated assessments of sensor and image quality that traditionally required extensive psychophysical testing with multiple human observers. It replicates frequency of seeing functions similar to those produced by human vision, allowing real-time assessment of multispectral sensors and fused imagery without human intervention.
Claims Coverage
The patent includes three independent claims covering a method, a system, and a program product for assessing optical sensor resolution through morphological orientation detection of Landolt C targets.
Automatic orientation detection using threshold adjustment and pixel tagging
The method obtains an image from an optical sensor, thresholds it to generate a binary image with contrasting colors using an initial threshold value, selects a target area, increases the threshold value, thresholds the target with the new value, tags pixels at the center, processes surrounding pixels tagging lighter colored pixels, determines orientation when a tagged pixel reaches the target edge, and repeats the process by decreasing the threshold if no edge tagging occurs.
Use of black and white contrasting colors and threshold value ranges
Utilizes black and white pixels as contrasting colors with threshold values between 0 and 255, wherein the initial threshold value is incremented by about 60 to generate the adjusted threshold, and orientation determination ceases if the adjusted threshold falls below a defined limit relative to the initial threshold.
Orientation determination via known pixel pattern comparisons for small target areas
For target areas sized 2 pixels by 3 pixels or 3 pixels by 2 pixels, compares the thresholded area to predefined sets of known pixel patterns representing orientations; if no match occurs, orientation is not determined.
System incorporating optical sensor and processor executing the orientation detection method
A system comprising an optical sensor, memory, and processor with resident program code that, when executed, performs the steps of obtaining an image, thresholding, selecting target area, thresholding with adjusted values, pixel tagging and processing, orientation determination, and repeated threshold adjustment.
Program product embodying the method for resolution assessment
A non-transitory computer readable medium containing program code that, when executed on hardware, obtains sensor images, performs adaptive thresholding and pixel tagging, processes pixels to detect edge contact, determines object orientation, and iteratively adjusts threshold values to conclude orientation assessment.
The independent claims collectively cover the inventive morphological method of adaptive thresholding and pixel tagging to automatically determine Landolt C orientation for sensor resolution assessment, implemented as a method, system, and program product with specific threshold adjustment and pattern matching techniques.
Stated Advantages
The morphological approach is more computationally efficient and accurate than contemporary methods.
Automated assessments emulate human visual frequency of seeing functions without requiring costly, laborious psychophysical testing by human observers.
The method enables real-time automatic assessment of individual sensors and fused multispectral imagery in terms of human visual perception, improving assessment efficiency and applicability.
Documented Applications
Assessing the resolution and digital quality of optical sensors such as visible, near infrared, short-wave infrared, and thermal sensors.
Evaluating images that have been algorithmically combined or fused from multiple sensors into synthetic sensors.
Supporting military and surveillance systems relying on digitally-based multi-spectral imaging for target detection and identification.
Interested in licensing this patent?