The Amplitude Co-occurrence Matrix Analysis Of Seismic Image And Its Application

Texture is one of the basic properties of the image. The co-occurrence matrix isone of the important methods for analysis and description of the image texture features.The seismic image is still a class of images, which could be extracted textureproperties by image texture analysis methods. Seismic image texture properties aredifferent from the conventional seismic attributes, because they have a certainstatistical significance. Amplitude co-occurrence matrix analysis is a commonly usedmethod for the seismic image texture feature extraction. This thesis first introduces theprinciple of amplitude co-occurrence matrix analysis method for image, and thendescribes the algorithm of seismic image texture analysis and program implementation.Finally, we describes the application effect for improving the vertical resolution ofseismic data and detecting faults using the seismic image texture analysis methods.Themain contents and results of this thesis are as follows:(1)Seismic image texture properties (energy property, contrast property,randomness property, homogeneity property), obtained by the two statistics fromseismic data, has clear physical meaning. After first statistic, amplitude co-occurrencematrix can be obtained from seismic image. The amplitude co-occurrence matrix is asymmetric matrix, and has relevant to the size of the seismic profile, the gray level,statistical step size, statistical direction. When applying the seismic image textureproperties, we should be based on the application purpose and actual seismic data, andselect the optimal size of the seismic profile, the gray level, statistical step size,statistical direction. After secondary statistic for amplitude co-occurrence matrix, wecould get the seismic image texture properties.(2)The seismic image texture energy properties can improve the verticalresolution of seismic data. Combining with the actual seismic data and its application purpose, we can select the best parameters, such as: the size of the seismic profile is9×7, the gray-scale size is16, statistics step is1, the statistics direction is the averagedirection of the four directions, and then extract seismic image texture energy propertyprofile for a survey line of the Longmenshan. Compared with the original seismic data,the energy property profile improves the vertical resolution, enhances with thecontinuity of the phase axis. Especially in the formation interface, it could get a moredetailed phase axis.(3)Dip scanning seismic image texture contrast property can improve theidentification of the fault. The seismic image texture contrast property has a goodadvantage, which is sensitive to the formation direction. When the statistical directionof the amplitude co-occurrence matrix is formation direction, and the quality seismicdata is very good, conventional seismic image texture contrast properties can detectfaults. However, we do not know the formation direction in advance, or the statisticdirections are not the strata direction, so the seismic image texture contrast propertiesat this time can not effectively detect faults. Therefore, we can calculate the formationdip and azimuth, construct the seismic profile in accordance with direction of the dip,and then calculate the seismic image texture contrast properties. In fact,we couldobtain the seismic image texture contrast property with dip scan, and they can improvethe ability to identify faults.