Shape description and retrieval using Conjugate Symmetric Sequency-Ordered Complex Hadamard Transform

Conjugate Symmetric Sequency-Ordered Complex Hadamard Transform (CS-SCHT) is a new version of Sequency-Ordered Complex Hadamard Transform. CS-SCHT is dyadic shift invariant and the spectrum has the conjugate symmetry property for real signals. This transform is very suitable to derive new shape descriptor because of its excellent properties. In this paper, CS-SCHT based shape descriptor (CS-SCHD) is proposed for image retrieval. In this method each sampled shape contour is expressed using a combined real signature which captures both local and global contour information. And then CS-SCHT is applied to the signature. The CS-SCHD is formed by the resulting transform coefficients to describe shape features. The image retrieval performance of the proposed descriptor is evaluated on the standard MPEG-7 shape database using standard measurement, and compared with three other transform based descriptors. The experimental results demonstrate that the proposed CS-SCHD reaches the highest precision at the same recall value among these descriptors, and show the effectiveness of the CS-SCHD.