基于优化熵函数二维最大熵阈值算法改进

Improvement of 2-D Maximum Entropy Threshold Algorithm Based on Optimal Entropy Function

根据熵函数在等概率场下取到最大值性质,对二维最大熵阈值法中的熵函数进行优化,将熵函数中对数和乘积运算简化为减法运算,得到目标函数具有意义明确、形式简洁、计算速度快的二维阈值自动选取方法。该法在保持二维最大熵阈值法对图像分割效果同时,又大大提高阈值选取速度,增强算法实时处理能力。这些结论正确性,既从理论推导过程中得到证明,也从对足迹图像分割的大量实验中得到验证。这种方法是保持二维最大熵阈值法分割效果不变、具有更快计算速度和实时处理能力、并在足迹图像分割中得到较好应用的一种自动阈值选取方法。

According to the property that the equivalent probability distributing has maximum entropy to get image threshold, this paper proposes an automatic threshing of two-dimensional maximum entropy approach, which optimizes entropy function of 2-D maximum entropy threshold by using subtraction instead of logarithm and multiplication. This approach gets the same result of image segmentation as 2-D maximum entropy threshold does and the computational time decreases at least by 15% on average. These results are proved theoretically and verified practically in a lot of footprint image segmentation experiments. This method of threshold automatic selection gets the same result of image segmentation as 2-D maximum entropy threshold does, but at a faster speed, therefore will find a wide application in footprint image segmentation.