Claims
- 1. A computer implemented method for relating disparate data models of a physical object to derive a first-range data model at a first-range distance, d, from a second-range data model at a second-range distance, d.sub.0, comprising the steps of:
- obtaining said second-range distance from signals transmitted from a sensor corresponding to said physical object;
- smoothing the second-range data model by a function of d and d.sub.0 ; and
- undersampling said smoothed second-range data model to obtain said first-range data model.
- 2. The computer implemented method of claim 1, wherein said first-range data model comprises a multi-dimensional first-range data model and said second-range data model comprises a multi-dimensional second-range data model.
- 3. The computer implemented method of claim 1, wherein said smoothing step comprises the step of masking said second-range data model by a masking operator, said masking operator having the same dimension as said second-range data model, said masking operator further having an equal number of odd elements in each dimension, said elements comprising values of a statistical distribution centered about a central coordinate of said masking operator, said number of odd elements being determined by a function of d and d.sub.0.
- 4. The computer implemented method of claim 3, wherein said masking operator comprises a two-dimensional square masking matrix having a first size and said second-range data model comprises a two-dimensional data matrix having a second size.
- 5. The computer implemented method of claim 4, wherein the first size of said masking matrix corresponds to the ratio of said first-range distance to said second-range distance.
- 6. The computer implemented method of claim 3, wherein said first size of said masking matrix is obtained from the equation m=1+2, wherein m is said size, said is a ceiling operator and said .sigma. equals a standard deviation of a multi-dimensional Gaussian statistical distribution.
- 7. The computer implemented method of claim 6, wherein the method further comprises the step of obtaining said .sigma. from the equation .sigma.=1/3((d/d.sub.0)-1).
- 8. The computer implemented method of claim 3, wherein said undersampling step comprises the step of extracting information from said second-range data model to generate said first-range data model having the same dimension as said second-range data model, said first-range date model being smaller in size in each dimension according to the difference in distance between said first-range distance and said second-range distance.
- 9. The computer implemented method of claim 4, wherein said step of undersampling comprises the step of using an undersampling factor, f, and further comprises the step of extracting from said smoothed second-range data model, having said second size of w-by z, those elements having row, i, and column, j, numbers less than w and z, respectively, and satisfying i=1+(a-1)f, and j=1+(b-1)f, where a,b are positive integers, said undersampling factor, f, further corresponding to the ratio of d to d.sub.0.
- 10. The computer implemented method of claim 1, wherein said first-range data model is a long-range data model and wherein said second-range data model is a short-range data model.
- 11. A method of operating an automatic target recognition system having at least one sensor, and a computer memory, comprising the steps of:
- creating a database of second-range data models in said computer memory;
- extracting an image of a potential target from said sensors;
- determining a distance, d.sub.0, between said sensors and said potential target;
- smoothing the second-range data model by masking said second-range data model by an m-by-m two-dimensional masking matrix, said masking matrix comprising elements comprising values of a Gaussian statistical distribution having a standard deviation, .sigma., and centered about a central coordinate of said two-dimension masking matrix, wherein said masking matrix dimension size, m, derives from the equation m=1+2��3.sigma.!! and said .sigma. equals the standard deviation of said two-dimensional Gaussian statistical distribution, and said .sigma. further deriving from the equation .sigma.=1/3((d/d.sub.0)-1;
- undersampling said smoothed second-range model to yield a first-range data model by extracting from said two-dimensional second-range data model information to generate said first-range data model having the same dimension as said second-range data model, said undersampling step using an undersampling factor, f, and further comprising the step of extracting from said smoothed two-dimensional second-range data model those elements having row, i, and column, j, numbers less than or equal to the size of the second-range data model matrix that satisfy the equations i=1+(a-1) f, and j=1+(b-1) f where a,b are positive integers, said undersampling factor, f, further equaling the ratio value of d to d.sub.0 ; and
- matching said first-range model against said extracted image to determine whether said potential target matches an object described by said model.
- 12. The method of claim 11, wherein said two-dimensional second-range data model comprises an absorptivity model.
- 13. The method of claim 11, wherein said two-dimensional second-range data model comprises a conductivity data model.
- 14. The method of claim 11, wherein said two-dimensional second-range data model comprises a range image data model.
- 15. The method of claim 11, wherein said two-dimensional second-range data model further comprises a thermal data model.
- 16. The method of claim 11, wherein said two-dimensional second-range data model further comprises a photometric data model.
- 17. A distance relating system in communication with an sensor system for generating first-range data models at a first-range distance from second-range data models at a second-range distance comprising:
- a memory for receiving and storing a plurality of said second-range data models;
- smoothing circuitry in communication with said memory for smoothing said stored second-range data models according to a function of said first-range distance and said second-range distance;
- undersampling circuitry in communication with said smoothing circuitry for undersampling said smoothed second-range data models in relation to said first-range distance to generate said first-range data models; and
- output circuitry in communication with said undersampling circuitry for outputting said first-range data model to said sensor system.
- 18. A method of operating an automatic target recognition system having at least one sensor and a computer memory, comprising the steps of:
- creating a database with target models at specified distances;
- extracting an image from said sensor;
- determining a distance, d.sub.0, between said target and said sensor;
- selecting a second-range model from said database, said second-range model having associated with it a distance d;
- using said distance d and d.sub.0 to construct a smoothing matrix;
- convolving said selected second-range model with said smoothing matrix thereby constructing a smoothed model;
- undersampling said smoothed model, thereby constructing a first-range model corresponding to distance d.sub.0 ; and
- comparing said first-range model and said image to determine if there is a match between said model and said image.
- 19. A computer implemented method for relating disparate data models of physical object to derive a first-range data model at a first-range distance from a second-range data model at a second-range distance comprising the steps of:
- obtaining said second-range distance from signals transmitted from a sensor corresponding to the physical object;
- smoothing the second-range data model by a function of d and d.sub.0 ; undersampling said smoothed second-range data model to obtain said first-range data model by function of d and d.sub.0 ; and
- transmitting said first-range data model as additional signals to said sensor and modifying said sensor in accordance with said first-range data model.
- 20. A computer implemented method for relating disparate data models of a physical object to derive a first-range data model at a first-range distance, d, from a second-range data model at a second-range distance, d.sub.0, comprising the steps of:
- converting physical characteristics of said physical object to signals by a sensor;
- obtaining said second-range distance from said signals transmitted from said sensor corresponding to said physical object;
- smoothing the second-range data model by a function of d and d.sub.0 ; and
- undersampling said smoothed second-range data model to obtain said first-range data model.
- 21. A method for relating disparate data models of a physical object to derive a first-range data model at a first-range distance, d, from a second-range data model at a second-range distance, d.sub.0, comprising the steps of:
- detecting physical characteristics of said physical object;
- converting said physical characteristic to signals by a sensor;
- obtaining said second-range distance from said signals transmitted from said sensor corresponding to said physical object;
- smoothing the second-range data model by a function of d and d.sub.0 ; and
- undersampling said smoothed second-range data model to obtain said first-range data model.
Parent Case Info
This application is a Continuation of application Ser. No. 07/510,761, filed Apr. 18, 1990, now abandoned.
US Referenced Citations (7)
Non-Patent Literature Citations (1)
Entry |
"Space-Scale Filtering", A.P. Witkin, Proceedings of the Eighth International Joint Conference on Artificial Intelligence, Aug., 1983, pp. 1019-1022. |
Continuations (1)
|
Number |
Date |
Country |
Parent |
510761 |
Apr 1990 |
|