Analysis of the parallel-sequential method of correlation calculations in the recognition problem
Abstract
The functional capabilities of the parallel-sequential method of correlation calculations are examined. Methods of correlation field formation are proposed which use, in a natural manner, a priori information on the class of the images being processed by increasing the 'weight' of the characteristic area of the reference image. These methods make it possible to increase the difference between the autocorrelation and cross-correlation maxima, thus increasing the reliability of threshold recognition. Experimental results are presented to validate the analysis.
- Publication:
-
Avtometriia
- Pub Date:
- April 1988
- Bibcode:
- 1988Avtme.......26B
- Keywords:
-
- Correlation Detection;
- Image Processing;
- Optical Data Processing;
- Sequential Analysis;
- Autocorrelation;
- Maxima;
- Parallel Processing (Computers);
- Optics