指针式仪表图像的快速分割法研究

在指针式仪表图像中，由于目标和背景的灰度比较单一，出现了直方图中大部分灰度级上像素数为零的现象。根据这一特点并且考虑到传统的熵最大法分割效果好但耗时的缺点，提出了一种改进的快速算法。由熵最大法的原理推理可知，对于像素数为零的灰度级可以不进行熵值的计算。因此该算法比传统的熵最大法多采用一个环节，用来判断灰度级上的像素数是否为零，如果为零则忽略对应的灰度级。这样既不会影响分割效果，又通过减少指数运算次数节省了处理时间。实验表明，该算法在不影响传统算法分割性能的基础上，提高了图像分割的效率，为指针式仪表图像的实时识别奠定了良好的基础。

Fast segmentation algorithm for image of gauge with pointer

No pixel exists on most gray-level in the histogram because both the object and the background have few gray-level in image of gauge with pointer. According to the phenomenon and taking the slowness of classical maximum entropy thresholding algorithm which does robust segmentation into account, an improved fast algorithm is proposed. On the gray-level that has no pixel, the value of entropy needn't be calculated based on the principle of the classical algorithm. So the improved algorithm adopts one more step in order to skip the calculating if no pixel exists on the gray-level. The improved algorithm does not only weaken the robustness at the same time but also reduces processtimeby decreasing the times ofexponent operation. Theresultofexperiments showsthatthe improvedalgorithm can improve the speed of image segmentation by reserving the robustness and lay the foundation for real-time image recognition of gauge with pointer.