动态结合全局阈值的图像分割方法,在图像灰度分布没有明显双峰及背景复杂的情况下也有很好的适应性。 用OTSU方法求得近红外图像的全局阈值s和各子图像的阈值d;利用双线性内插值算法求得与原图像 等大的阈值矩阵D。通过调整比例系数k,最后完成阈值矩阵T的构造和图像分割。此方法能有效分割出石化炼油炉的炉管图像。

In the case of image gray distribution without significant double peaks and the complex background, the dynamic image segmentation method combined with global threshold has good adaptability. The global threshold of near-infrared image s, and the threshold value of the small-block images d are evaluated by the method of OTSU, and the threshold matrix D with the original image by the bilinear interpolation algorithm. Finally, the structure of threshold matrix T and segmentation of image are completed by adjusting the scale coefficient k. This method can effectively split images of the furnace tube in petrochemical refinery.