Implementation of Objects Recognition in Seismic Image via Artificial Neural Network (ANN)

• Main Author: Wan , Chin Ee
• Format: Final Year Project
• Language: English
• Published: UNIVERSITI TEKNOLOGI PETRONAS 2012
• Online Access: http://utpedia.utp.edu.my/6341/1/Wan%20Chin%20Ee_12893_Dissertation.pdf; http://utpedia.utp.edu.my/6341/

Seismic image processing is necessary in oil and gas exploration to identify the existence of potential reservoir by classifying the seismic image into different sections. These sections, also known as objects made up of different patterns which portraying the structure of subsurface. This project aims to develop a data mining algorithm embedded in a system that has ability to recognize the objects of channel and fault in seismic image. The method chosen is artificial neural network (ANN) which consists of input layer, hidden layer and output layer. Each layer is made up of numbers of neuron nodes to receive input data from preceding layers and output value to next layer until final output is determined from output layer. The ANN is trained and tested via MATLAB Neural Network Pattern Recognition Toolbox (nprtool) and MATLAB Neural Network Toolbox (nntool). 2-dimension (2D) seismic image is converted into gray scale image via MATLAB Image Processing Toolbox (imtool) and Grey-level co-occurrence matrix (GLCM) which serve as input to the ANN is retrieved from the gray scale image. Result is displayed by the system informing user whether the input image is channel, fault or neither both.