最小误差分割算法的图像分割性能优异,但一维的最小误差分割算法容易受到噪声的干扰。利用图像的二维直方图,二维最小误差分割算法不仅能够利用图像的灰度信息,同时利用了相邻像素之间的邻域信息,取得更加理想的分割效果。但在实际使用的过程中,二维最小误差算法采用穷尽搜索的算法运算时间长,二维直线型最小误差分割算法无法反映全局最优解,降维形式的最小误差算法复杂度高。本文将结合粒子群优化算法(PSO) 将二维最小误差分割算法应用在红外图像上,大大提升了算法的求解速度,能够在实现更低对比度的红外图像分割的同时满足工程中实时检测的要求。

The minimum error segmentation algorithm for image segmentation performs excellently, but the minimum error segmentation algorithm of one-dimensional is susceptible to noise. By using two dimensional histogram of the image, two-dimensional minimum error segmentation algorithm can not only use the gray value of the image, but also the adjacent pixels information, and obtain more ideal segmentation effect. Unfortunately, two dimensional minimum error algorithm of exhaustive search wastes a lot of time, two linear dimensional minimum error segmentation algorithm can not reflect the global optimal solution and the way of dimension reduction is complicated. This paper will combine the particle swarm optimization algorithm(PSO) to the two-dimensional minimum error segmentation algorithm and apply it into infrared images to greatly enhance the speed of algorithm in infrared image segmentation. A lower signal-to-noise ratio was achieved, and the requirements of real-time detection of engineering was met.