Digital image watermarking requires the embedded watermark to be resilient and retrievable even if the watermarked content undergoes one or more image processing operations, in part or as a whole. Consequently, it is proposed to evolve the watermark embedding strategy to satisfy the conflicting objectives of causing image content changes that are imperceptible to the human eye while being extremely robust against intentional and unintentional image processing attacks. It is assumed that the attacker knows the algorithm and so, the performance and security depend on the algorithm itself and the key(s) used. The attacker is expected to exploit the specific weakness of the algorithm to defeat the watermarking.