In this thesis, a method for color image watermarking is suggested. It requires the original image for watermark extraction. This method exploit the discrete wavelet transform features, where the large coefficients in high frequency sub bands are used as host place to embed the invisible fractal watermark image using the Haar transform to embed the huge data. The proposed system exploit the fractal image technique using the midpoint displacement method fractal technique (By using the H-dimension as entered real number) to more secure and preserve the watermark from intentional or unintentional attacker. The system can be used for owner identification, proof of ownership, and transactional watermarks (fingerprinting) also it is capable to survive against the high degree of compression produced by JPEG2000. By using Haar transform we can embed the fractal watermark image with modulation factor value to satisfy robustness against compression. The fractal watermarking image was fully extracted under degree of compression that makes the size of image decrease into (1:3) of the original size. In spite of this decreasing in the size of image the distortion bits is very little or underprivileged and the image is capable to survive the extraction. To be practical, the system was tested by using various images evaluated by adopting many fidelity measurement, the test results indicated that the distortion bits is increased when the compression size decreases, and the quality of image also decreases. The system was incomplete and robust watermark, then it considered as a private watermark. By using midpoint displacement method can generate image of any size with mean equal to zero and standard deviation equal to one. The proposed system uses RGB color space.