模擬退火參數偵測系統於物件偵測與震測圖形之應用(III)

Abstract

類神經網路的理論及應用之研究，近年來在國際上愈來愈重要，應用範圍也相當的廣泛。我們提出利用以機率為基礎的模擬退火演算法的圖形偵測系統，偵測直線、圓、橢圓、與雙曲線的圖型參數。不同於赫夫轉換類神經網路(Hough transform neural network)會受到收斂到區域最小值的影響，此一系統利用模擬退火演算法求得的圖型參數，會使得誤差為全域的最小值(最佳化)，因而能夠使得偵測的直線、圓、橢圓、與雙曲線的參數將更精確。在實驗上，此系統於影像中圖形偵測成功後，我們將應用於震測圖形中的直線的直接波與雙曲線的反射波的參數之偵測，其震測圖形包含模擬震測圖形與實際震測圖形，偵測後的結果將有助於震測訊號的解釋。在更進一步的應用上，有別於傳統的速度分析方法，我們將用於 Common Depth Point (CDP) gather 的 data 作雙曲線的參數偵測，由偵測到的參數，求出震波的 root-mean-squared 速度，再對震測圖作 normal moveout correction (NMO) 修正，及 stacking，得出stacked 後的震測資料，以反映地層的震波的垂直來回時間的震測圖形。

Recently the development of theory and application of neural networks become increasingly important in international community. And the areas of the applications are quite wide spread. We propose the pattern detection system using probability-based simulated annealing that can detect the parameters of the lines, circles, ellipses, and hyperbolas. The system can not only detect parameters with a higher precision, but also a globally optimal solution rather than a local optimal solution by the Hough transform neural network. After the success of the system in image pattern detection, we will apply it to detect the parameters of the line of direct wave and the hyperbola of reflection wave in the simulated one-shot seismogram and real seismic data. The detection results will improve the seismic interpretations. On the further seismic data processing, different from the conventional velocity analysis, we will apply the system to detect the hyperbolic patterns on common depth point (CDP) gather data. The root-mean-squared velocities can be solved by the parameters of the detected hyperbolas and used in the normal moveout correction (NMO) and stacking. Then we will obtain the stacked seismogram for the vertical travelling time of the geologic model.