基于灰度直方图拟合曲线的数字图像多阈值分割技术研究

Multi-threshold dividing technology of digital image based on gray histogram fitting curves

在数字图像处理中,图像分割技术能够对待处理部分进行分割处理,以突出显示所要研究的目标对象。在对数字图像灰度直方图进行高次样条函数曲线拟合之后,通过求取若干个局部最小值作为图像的分割阈值,对图像像素进行多阈值分割,最后采用区域背景像素点增长合并法进行背景区域合并,从而获得完整清晰反映图像各个组成部分的分割结果。实验结果证明,此方法能准确表达图像各个灰度组成部分,克服了传统单阈值分割所带来的缺陷与不足。

In digital image processing,image-dividing technology can divide and display research target. After interpolating digital image gray histogram curves by high order spline function, several local minimum values are gotten and used as image＇s dividing threshold. Through multi-dividing of image, pixel growing and merging technology is used to merge image＇s background region, thus clear and complete dividing image result is obtained. Experimental results prove that this technology can display image＇s dif- ferent gray elements,thus flaw and deficiency of traditional single threshold image dividing technology can be overcome.