多策略融合鲸鱼算法与二维最大熵的图像分割

Multi-strategy Fusion Whale Algorithm and Two-dimensional Maximum Entropy Image Segmentation

传统的图像分割算法存在抗噪声差、迭代速度不高等缺陷,为了提高图像分割的质量,提出一种引入反向学习策略的鲸鱼优化算法(WOA)与二维最大熵结合的图像分割方法,并通过结合Sobol序列、自适应权重系数、收敛因子的非线性调整优化后的WOA算法,得到目标图像的最优阈值,并对其进行分割。通过与原始WOA算法在测试函数上的对比,表明改进算法具有较好的收敛性和较快的收敛速度。将改进算法实际运用到草坪背景下,分割后图片的峰值信噪比值相较于另外2种对比算法分别提高了5.2%、3.5%,在天空背景下分别提高了3.6%、2.2%,证明了改进后的算法可以提高分割的图片质量,体现了该算法的优越性。

Because the traditional image segmentation algorithm had the defects of poor noise resistance and low iteration speed,in order to improve the quality of image segmentation,a whale optimization algorithm(WOA)with reverse learning strategy and a two-dimensional maximum entropy image segmentation method were proposed,and the WOA algorithm was optimized by nonlinear adjustment combining Sobol sequence,adaptive weight coefficient and convergence factor.The optimal threshold of the target image was obtained and segmented.Compared with the original WOA algorithm in the test function,the improved algorithm has better convergence and faster convergence speed.When the improved algorithm is applied to the lawn background,the peak signal-to-noise ratio value of the segmented image is increased by 5.2%and 3.5%respectively compared with the other two contrast algorithms,and increased by 3.6%and 2.2%respectively under the sky background.The improved algorithm can improve the quality of the segmented image,reflecting the advantages of the algorithm.