Protection, management, and indexing of digital contents are becoming more prominent with the increasing popularity of electronic commerce and on-line services. Multimedia watermarking has received increased attention with regard to these problems. This dissertation presents content-based image watermarking method that is resilient against geometric transformations. Since it is quite easy with modern computers to apply geometric transformations to image, there is an obvious necessity for geometric-distortion resilient watermarking. Although there are some approaches dealing with the geometric distortion, geometric transformations with cropping, random bending attack, non-isotropic scaling, and projective transformation still remain as difficult problems. The remained problems are addressed by using refined pattern recognition techniques.