Higher Moment Image Features Computation
Original Publication Date: 1986-Aug-01
Included in the Prior Art Database: 2005-Mar-09
This invention relates to the computation of moments in binary image feature extraction. In such extraction the array location of the pels, for example, which border an object, rather than the pel values themselves, are of interest. Various properties may be tested, such as skewness, by way of higher order moments. The higher order moments involve polynomial representation. Further, second and third order moments are computationally most frequently involved. Since the precision width of the numbers i bits, in contrast to the commercially available 16-bit width of single chip multipliers, computational avoidance is of advantage. One critical observation was that the second order moments of consecutive points could be calculated by adding and shifting a linear increment to a current value to obtain the next moment value.