检测井壁图像上平面地质特征的改进霍夫变换

Improved Hough transform to detect planar geologic features in borehole images

为实现井壁图像上平面地质特征的自动提取,研究了图像中单周期正弦曲线的检测方法。提出一种改进的霍夫变换,该方法基于正弦曲线上的三个相关联点在二维累加器中进行投票,从而确定相位与基线深度;然后在一维累加器中投票,确定振幅。对模拟图像及井壁图像进行了处理,并与传统霍夫变换、快速霍夫变换进行了对比。结果表明,该方法不仅提高了检测速度,而且具有较高的准确率。

For the implementation of automatically extracting planar geologic features in borehole images, the methods of detecting single cycle sinusoidal curves in images were studied. An improved Hough transform were proposed. The proposed method voted in the two-dimensional accumulator based on three related points in the sinusoidal curve, to determine the phase and baseline depth. Then vote was performed in one-dimensional accumulator, to determine the amplitude. The simulated images and borehole images were processed. The proposed method was compared with the traditional Hough transform and fast Hough transform. The results show that the proposed method not only improves the detection efficiency, but also has high accuracy.