Digital image watermarking is the insertion of watermarks into digital image media. Several types of watermarking methods used are Discrete Cosine Transform (DCT) and Discrete Fourier Transform (DFT). Both of these watermarking methods work in the frequency domain (transform). Digital image watermarking using the frequency domain is carried out on the frequency coefficient. This study used 30 digital image data as material for digital image watermaking analysis with 10 data each in binary, grayscale and color digital images in jpg, png and bmp formats. Digital images in the binary and grayscale domains are conversions from digital images in the true color (RGB) domain. Digital image watermarking includes three main processes, namely embedding the watermarked image on the original digital image, extracting the watermarked image and measuring the correlation between the two digital images. Correlation aims to measure two variables that have the same relationship. The technology used in this research work is MATLAB (Matrix Laboratory) as a high-performance programming language for computing in solving problems with solutions expressed in mathematical notation. The results of the discussion prove that the watermarking process in terms of color, for DCT, RGB is better and binary is better for DFT. And the watermaking process, in terms of the type of watermark inserted, for both DCT and DFT, a good watermark is an invisible watermark.