Studies how to embed information into image using public watermarking, where bit can be extracted without original image. Turbo code was used for bit error correction. Image restoration technique was used to estimate the orginal image for extracting information bits. Information bits were embeded in DCT domain. This research also investigated the method to determine the optimum range of zigzag DCT coefficients to be embedded for each image at the same visibility level. The watermarking robustness for various image processing methods was tested and the maximum embedded bits at 10 -3 bit error rate was found. The turbo code can increase the maximum number of embedded bits more than 3 times when compared to the cases of no coding and using Hamming code (7, 4) instead of Turbo code. This is because the turbo code is more efficient at low SNR level and image watermarking is usually under this condition.