Coding of contour data using DCT transform

The efficiency of the DCT transform in the coding of contour data is presented. Test contours extracted from gray level or binary images are represented by two one-dimensional sequences x(i) and y(i) representing the Cartesian coordinates of the boundary points of the contour. The contour sequences representing the contour data are combined to form one sequence. The new sequence is then divided into a number of sub-arrays, which are transformed using one-dimensional DCT transform. The spectral coefficients having the largest variances are then coded and transmitted or stored. Comparison of the coding results using different transforms, with respect to mean square error versus the compression rate, are reported. It is shown that the DCT transform has a better coding efficiency than other transforms.