一种新的自适应工程图像分割算法

为满足水厂矾花实时检测系统中对矾花图像分割的要求，分析了工程图像的常用阈值分割算法，并针对其不足提出了一种新的基于直方图分析的自适应工程阈值分割算法；算法能根据直方图类型灵活处理以提高自适应能力，并利用最大方差法和直方图类型确定分割阈值以减小阈值偏差，克服了局部阈值分割方法和最大方差法中阈值确定的不足，提高了系统的处理速度和鲁棒性能；在水厂自动加矾系统实时矾花检测系统的实际应用中取得了良好的效果。

Self-adaptive Threshold Segmentation Algorithm About Engineering Images

For the purpose of alum image segmentation in waterworks, a self-adaptive threshold segmentation algorithm based on histogram is presented after analysis of the shortage of current threshold segmentation algorithm. The new algorithm can deftly handle the type of histogram so as to improve self-adaptability, and reduce the threshold error by confirming the value of threshold with the Maxium Variance algorithm and the type of histogram. This algorithm avoids the shortages of local threshold algorithm segmentation and Maximal Variance algorithm, so that the process speed and the robustness are increased obviously. It has triumphantly used in measuring alum images system in waterworks.