【摘要】 针对已往阈值分割算法对灰度图像的处理只利用其灰度属性,未考虑像素的空间位置关系,致使分割结果不连续以及不准确的问题,提出了一个新的自适应阈值分割算法。该方法通过结合像素空间几何距离与灰度值关系,运用模糊理论处理不确定因素的优势,充分利用分类集合内数据的聚集程度和分类集合间的间距的影响,保证了图像分割结果的准确性;借助分类集合的均值和方差,设计了新的最佳阈值判别标准。同已有阈值分割算法相比,该算法具有较好的稳定性和鲁棒性,对于灰度直方图呈单峰的多数图像能得到较好的分割效果。更多还原

【Abstract】 Aiming at solving the problems that most thresholding segmentation algorithms for grayscale image processing only take the gray-scale property into consideration without investigating the spatial relationship of the pixels,which lead to discontinuous and inaccurate of the segmentation results,an novel automatic thresholding algorithm based on the fuzzy theory is proposed.The new method,which utilizes the advantages of combination of spatial geometry distances and gray-scale property and considers the distances of between-class and within-class,ensures the accuracy and continuity of the image segmentation.Furthermore,a new criterion for threshold selection is also proposed with the mean and variance of different classes in mind.Experimental results demonstrate that the method has good stability and robustness compared with other recent proposed methods in processing vagueness images with unimodal histograms distributions.更多还原