提出了一种新的基于细菌趋药性(BC)算法的盲图像分离方法,利用图像信号的规范四阶累积量作为目标函数,使用BC算法对目标函数进行优化以实现图像的盲分离。每分离出一幅图像后,从混合图像中消除该幅图像成分后再进行下一次分离,从而最终实现所有源图像的逐次分离。仿真结果表明,本文算法能够有效实现对多幅混合自然图像的盲分离,且具有较好的分离效果。 A new blind image separation method based on bacterial chemotaxis (BC) algorithm is proposed. By using the standard fourth-order cumulant of image signal as objective function, the BC algorithm is used to optimize the objective function to achieve blind separation of images. After each image is separated, the image component is removed from the mixed image and then separated for the next time, so as to finally achieve the successive separation of all the source images. The simulation results show that this algorithm can effectively realize the blind separation of many mixed natural images and has a good separation effect.