Statistical optimization of radial vectors for pattern reconstruction/retrieval

Abstract

The size of radial vector for a given pattern increases as the pattern size increases due to increased perimeter of pattern under test. The pattern can be ideally reconstructed using all the radial vectors. However, the vector size may be optimized in order to reconstruct the pattern of the same quality as in an ideal case. In the presented work, the radial vector size is optimized using the statistical analysis of radii profile based on standard deviation, area and perimeter. The reconstructed pattern is approximated to its maximum towards the original pattern by maintaining the standard deviation, area and perimeter. The radii profile in each quadrant is used to get the extremes and figure aspect of the pattern. All the radii are computed about the centre of mass of the pattern under test.