Very large size image databases are being built for different applications, such as, digital library, museum object management, individual picture and photograph collections. With the fast development of INTERNET and mass storage devices, these image archives are made publicly accessible, and there is an increasing demand on techniques for searching and retrieving the images from the databases. The classical method using keyword of an image has been out of date due to substantial manual work and therefore content-based image indexing methods have become popular in the last decade. Latest research on image indexing concentrates on the techniques in compressed domain in order to reduce the storage cost because of large image database size. In this thesis, the indexing techniques from the classical text-based to the latest compressed domain content-based are critically reviewed. It is widely known that the wavelet-based techniques have a high potential to provide a superior coding and indexing performance in the compressed domain. In this thesis, several indexing techniques in the wavelet domain are presented. First, two techniques in the embedded zerotree wavelet framework are proposed. These techniques are based on the histogram of the number of significant wavelet coefficients. Four indexing techniques in the JPEG2000 framework are then proposed. These techniques primarily based on bit-plane and the packet header of JPEG2000 bit-stream. Experimental results show that the proposed techniques can achieve retrieval efficiency up to 95%. These techniques can be integrated easily with the recently established MPEG-4 and JPEG2000 standards.