针对图像全局Otsu分割算法和传统的局部阈值分割算法在复杂背景图像分割中的不足,提出了一种采用局部动态阈值的图像分割算法。首先通过分析图像直方图类型对图像进行划分,并使用Otsu算法确定各个子图像的阈值。然后对阈值矩阵进行平滑和插值处理,使之成为和原图像像素数目相等的新阈值矩阵。最终利用此阈值矩阵完成图像分割。实验结果表明,该算法对复杂背景图像的分割效果好于全局Otsu分割算法和传统局部阈值分割算法,是一种有效的分割算法。

For the disadvantage of global Otsu algorithm and traditional local threshold segmentation method used in the complex background image segmentation, a local dynamic threshold image segmentation algorithm is proposed. The image was firstly partitioned into sub-images by analyzing the image histogram type, and the threshold was calculated for each sub-image with the Otsu algorithm. After using the methods of smoothing and interpolation in the threshold matrix, the threshold matrix transformed to a new threshold matrix which had amount of pixels of origin image, completing the image segmentation. Experiments show that this algorithm of image segmentation in complex background is better than the global Otsu algorithm and the traditional local segmentation algorithm, proving an effective segmentation algorithm.