Abstract: In order to optimize the fidelity and robustness，this paper transforms the digital image watermark into a multi-object problem. Aiming at the problem that the traditional artificial bee colony algorithm is slow convergence speed and easy to fall into local optimal solution，an improved artificial bee colony algorithm is proposed and applied to the digital image watermark processing. The algorithm takes the food source intensity factor and the neighborhood factor into account to update the food source and improve the convergence speed. It uses the DCT to decompose the original image，embeds the watermark into the original image and extracts the watermark blindly. Experimental results show that，the algorithm has a good fidelity and robustness.

Key words: watermark, artificial bee colony, convergence, fidelity, robustness