中文说明： 基于二次最大类间方差法的自适应canny算子，采用Canny算子进行边缘检测时，梯度图像需要进行模非极大值抑制，然后求取双阂值提取边缘，但目前双阈值的求取无法避免人工设定的影响，试验表明，针对不同的图像采用相同的闽值，边缘检测效果差异很大．这一点限制了Canny算子在实际中的应用．针对这一问题，提出基于梯度幅度直方图和类内方差最小化自适应的确定高低 阌值的方法，可针对不同的图像，实现双闽值的自适应提取，不需要人为设定任何参数，采用模糊控制技术提取边缘 像素，实验结果证明了该算法的有效性．
English Description：
Based on the adaptive Canny operator of quadratic maximum variance between classes, when Canny operator is used for edge detection, gradient image needs to be suppressed by modulus non maximum, and then double threshold is obtained to extract edge. However, the current double threshold can not avoid the influence of manual setting. Experiments show that the effect of edge detection varies greatly with different images using the same threshold To solve this problem, an adaptive algorithm based on gradient amplitude histogram and within class variance minimization is proposed