为克服优化算法易陷入局部最优和收敛速度慢的局限,提高扩展膜系统在图像处理领域的优化性能,提出一种基于扩展型活性膜系统(P system)的改进北方苍鹰优化(improved northern goshawk optimization,INGO)算法——PINGO.采用北方苍鹰优化算法作为基本膜中的进化规则,通过更新苍鹰的状态进化基本膜中的对象,将INGO算法作为局部进化规则来进化子膜中的对象.该系统根据活性膜自身的特点在基本膜中溶解或产生子膜,通信规则用于实现不同膜之间的信息交换与共享,避免算法陷入局部最优.在数据集BSD300和BSD500上,分别采用海鸥优化(seagull optimization algorithm,SOA)算法、灰狼优化(grey wolf optimizer,GWO)算法、INGO算法和PINGO算法,对不同优化阈值个数的图像进行分割.结果表明,PINGO算法在分割后的图像上的峰值信噪比均优于其他算法,特征相似度最优值也占了83%,在保持色彩与纹理的同时提高了分割的准确性.研究结果表明了所提彩色图像分割方法的有效性.
Color image segmentation method based on extended active membrane system
To overcome the limitations when applying optimization algorithm,such as being prone to local optima and low convergence speed,and enhance performance of extended membrane systems in image processing,an improved northern goshawk optimization(INGO)algorithm,called PINGO,based on an extended active membrane system(P system),is proposed.In PINGO,the northern goshawk optimization(NGO)algorithm serve as the evolution rule in the basic membrane,to evolving objects by updating the goshawk's state,while the improved NGO acts as a local rule for sub-membranes.The system generates or dissolves sub-membranes in the basic membrane according to its characteristics.Communication rules enable information exchange between membranes in order to avoid local optima.The proposed segmentation method PINGO and other comparative algorithms,including seagull optimization algorithm(SOA)、grey wolf optimizer(GWO)and INGO,is evaluated on BSD300 and BSD500 datasets for different images with different number of optimize thresholds.The segmented images of PINGO achieved the best peak signal to noise ratio and 83%best feature similarity maxima,improving segmentation accuracy while maintaining color and texture.Experimental results demonstrate the effectiveness of the proposed method.
国家科技期刊平台
NSTL
万方数据
