基于改进广义Hough变换的高效测井图像校正

Well log curve images skew correction based on improved generalized Hough transform

在图像处理和计算机视觉中,广义Hough变换可以检测任意形状的曲线,而且较Hough变换运算量小,存储空间要求小。但是对于测井曲线这种大图像用广义Hough变换寻找倾斜角在速度上仍显迟钝。提出了一种改进的广义Hough变换,只选边界线段的端点作为参考点,可以较快地检测出倾斜角,速度比广义Hough提高了10倍以上。在图像校正方面,又提出了一种新的改进逆旋转算法,能有效地解决孔洞和锯齿现象。与已有的算法比较,新算法更快、更精确、消耗的内存更小。运算时间只有传统算法的一半左右。

In the field of the image processing and computer vision the Generalized Hough transform can examine various types of curves and the calculation amount and memory in demand is relatively less than the original Hough transform.For the big images as the well log curve image the calculation speed based on the generalized Hough transform still seems slow.An improved generalized Hough transform is presented,which can checks and measures the image angle at a fast speed by setting reference points on the shape edge＇s ending point.The calculation speed increases by more than 10 times than the generalized Hough transform.In the view of the image skew correction an improved image athwart rotation is proposed.It can solve the problems of the holes and sawtooth.Compared with the previous methods this new arithmetic is much faster,more accurate and requires smaller memory.Besides it almost saves half of the computation time.