基于结构导向的梯度属性边缘检测技术

Edge detection technique based on structure-directed gradient attribute

地震数据中的不连续性检测对于识别和刻画地下地质体边界具有重要意义.随着勘探目标向岩性油藏和复杂断块油藏的转变,解释人员需要对小断裂、微裂缝等微幅地质体边界信息进行精细刻画.常规边缘检测算子由于算法的局限性不能满足上述需求.本文通过引入能量归一化和沿层检测思想对常规检测算子进行改进,提出一种基于结构导向的梯度属性边缘检测技术.该技术通过结构张量估计得到局部地层区域的倾角和方位角,利用得到的倾角和方位角信息作为约束,使用二维边缘检测算子实现三维数据中的沿层不连续性检测.经过实际资料处理,并与结构导向方差属性对比分析,证实该方法在地层不连续性检测方面具有很好效果,检测得到的不连续性信息在细节上表现更丰富,可以作为地层横向差异和地质体边界识别的有力工具.

It is important to detect the uncontinuity of seismic data for the identification and depiction of underground abnormal geologic bodies. With the exploration target turning to lithologic reservoirs and complex fault block reservoirs, it is necessary to precisely depict the edge of micro geologic bodies such as small fault and micro-crack. It cannot be realized by using conventional edge detection operator because of its algorithmic limitation. We improve the operator through introducing energy normalization and detection along layers, and propose an edge detection technique based on structure-directed gradient attribute. In this technique, we use structure tensor to estimate the dip and azimuth of some local region, and then use them as constraints. Thus the uncontinuity detection along layers in 3D seismic data with 2D edge detection operator is realized. After field data processing and the comparison with structure directed variance attribute, this method demonstrates its effectiveness in uncontinuity detection, and the derived uncontinuity information shows plenty of details. Thus it can be very useful inthe identification of horizontal difference of layers and the edge of geologic bodies.