动态阈值结合全局阈值对图像进行分割

Image Segmentation Using Dynamic Threshold Combinedwith Global Threshold

首先讨论了图像阈值分割的两种方法 :全局阈值图像分割和动态阈值图像分割 .当背景的灰度值可看作恒定时 ,使用全局阈值一般会有较好的效果 ;在许多情况下 ,背景的灰度值并不是常数 ,物体和背景的对比度也有变化 ,我们可将一个随图像位置缓慢变化的函数设为阈值 ,即采用动态阈值 .但由于我们要对图像分块 ,简单的动态阈值分割所得的图像通常包含人为的边界和阴影 ,为改善图像分割效果 ,本文提出了把全局阈值和动态阈值结合起来对图像进行分割的一种新的图像阈值分割方法 .实验表明 ,该方法比单独采用全局阈值或动态阈值方法性能优越 .

The paper firstly discussed two ways for image segmentation:image segmentation by global threshold and image segmentation by dynamic threshold.If the background of an image is constant throughout the image,a global threshold may work well.In many cases,the background gray level is not constant,and the contrast of an object varies within the image.In these cases,we may use a threshold which is a slowly varied function of position in the image,that is,use a dynamic threshold.Because we must divide the image into sectors,the image divided by simple dynamic threshold usually includes shadow and man-made boundaries.To improve the effect of image segmentation,the paper introduced a new way for image segmentation which using dynamic threshold combined with global threshold.Examinations indicate that this way works better than the other two ways.