摘要: |
针对位场数据处理中边界定位精度和i只别能力的问题,提出一种新的边界检测方法一均方差比归一化垂向 梯度法。均方差比(MSER)可进行边界识别,是针对边界点异常方向性和均方差衡1:数据波动性提出的,对全K数 据点4个方向的均方差进行归一化后选择各个数据点均方差比的最大值作为滤波输出实现的;均方差比归一化垂向 梯度(NVD-MSEK)可进行边界增强,是通过均方差比的垂向梯度及it总梯度的比值实现的。模型试验对比分析结 果表明,NVD-MSER方法具有计算稳定性强、反映的边界位置连续性好、与实际模型边界对比偏差小的优点,这说明 NVD-MSER法有较强的边界检测能力。用NVD-MSER法可以检测出黑龙江虎林盆地19条断裂,而欧拉反褶积只能 识别出11条断裂,说明NVD-MSER法增强了对断裂平面位置的识别能力。 |
关键词: 方位检测 边界检测;归一化;均方差比;梯度 |
DOI:10.3969/j.issn.1673-5005.2012.02.014 |
分类号:P312; P63 |
基金项目:同家然科学重点基金项R (40739905);闰家油气选区.项目(14B09XQ1201) |
|
Edge detection of potential field based on normalized vertical gradient of mean square error ratio |
WANG Yan-guo, WANG Zhu-wen, ZHANG Feng-xu, ZHANG Jin, TAI Zhen-hua, GUO Can-can
|
(College of Geo-Exploration Science and Technology, Jilin University, Changchun 130026, China)
|
Abstract: |
Aimed at the problems of edge positioning precision and recognilion capability in potential-Hield data processing, a new edge detection technique based on the normalized vertical derivative of mean square error ratio (NVD-MSER) was presented. Mean square error ratio ( MSER) can be used for edge recognition, and it is based on the directionality of boundary anomaly and ihe data volatility evaluated by mean square error. MSKR can be achieved by the following steps : normalizing the four-directional standard deviation of the whole region,then selecting maximum from four directional MSERs of each and every data. NVD-MSER can be used for edge enhancement and it is the ratio of the vertical gradient of MSKR and the total gradient. Comparative analyses of model test show that, the NVD-MSER has strong computing stability, good recognized edge continuity, and small deviation with model edge, so it has strong detection capability. In the application, NVD-MSER can detect 19 faults in Hulin Basin of Heilongjiangt while Euler deconvolution can identify 11 faults. It is shown that the NVD-MSER method can enhance recognized stability in the planar location of faults. |
Key words: azimuth detection edge detection normalization mean square error ratio gradient |