System and method for target separation of closely spaced targets in automatic target recognition
The Target Separation Algorithms (TSAs) are used to improve the results of Automated Target Recognition (ATR). The task of the TSAs is to separate two or more closely spaced targets in Regions of Interest (ROIs), to separate targets from objects like trees, buildings, etc., in a ROI, or to separate targets from clutter and shadows. The outputs of the TSA separations are inputs to ATR, which identify the type of target based on a template database. TSA may include eight algorithms. These algorithms may use average signal magnitude, support vector machines, rotating lines, and topological grids for target separation in ROI. TSA algorithms can be applied together or separately in different combinations depending on case complexity, required accuracy, and time of computation.
1. An image processor comprising:
a sensor for sensing and recording a single two-dimensional (2-D) image as 2-D image data, the 2-D image data comprising 2-D pixel locations and corresponding signal magnitudes, one of the signal magnitudes for each of the 2-D pixel locations;
an automated target cueing component for identifying one or more regions of interest in the single 2-D image;
a target separation component for separating the 2-D image data in the respective regions of interest; and
an automated target recognition component for creating target information corresponding to the single 2-D image,
wherein for each region of the regions of interest the target separation component is configured
to compute an average signal magnitude as a weighted average of a minimum and a maximum of the respective signal magnitudes of the 2-D pixel locations of the 2-D image data within the region,
to compare the average signal magnitude to the respective signal magnitudes of the 2-D image data within the region, and
to distinguish the 2-D image data within the region based on the average signal magnitude.
2. The image processor of claim 1 wherein the target separation component is further configured to separate objects from clutter in one of the regions of interest using the average signal magnitude.
3. The image processor of claim 1 wherein the target separation component is further configured to separate two objects in one of the regions of interest by using the average signal magnitude.
4. The image processor of claim 3 wherein the target separation component is further configured to separate three or more objects in the one of the regions of interest by using the average signal magnitude.
5. The image processor of claim 1 wherein the target separation component is further configured to filter the 2-D image data distinguished by the target separation component.
6. The image processor of claim 1 wherein the target separation component is further configured to restore the 2-D image data distinguished by the target separation component.
7. The image processor of claim 1 wherein the target separation component is further configured to use support vector machines to separate objects in the respective regions of interest.
8. The image processor of claim 1 wherein the target separation component is further configured to use rotating lines about or near a center of mass to separate objects in the respective regions of interest.
9. The image processor of claim 1 wherein the target separation component is further configured to use horizontal or vertical grids to separate two objects in the respective regions of interest.
10. The image processor of claim 1 wherein the target separation component is further configured to use linear or nonlinear borders to separate two or more objects in the respective regions of interest.
11. A method of target separation comprising:
sensing and recording a single two-dimensional (2-D) image as 2-D image data, the 2-D image data comprising 2-D pixel locations and corresponding signal magnitudes, one of the signal magnitudes for each of the 2-D pixel locations;
identifying one or more regions of interest in the single 2-D image;
separating the 2-D image data in the respective regions of interest; and
recognizing one or more targets and creating target information corresponding to the single 2-D image,
wherein for each region of the regions of interest the separating of the 2-D image data comprises:
computing an average signal magnitude as a weighted average of a minimum and a maximum of the respective signal magnitudes of the 2-D pixel locations of the 2-D image data within the region;
comparing the average signal magnitude to the respective signal magnitudes of the 2-D image data within the region; and
distinguishing the 2-D image data within the region based on the average signal magnitude.