Research And Implementation For Geological Image Processing Algorithms

Image processing is provided with quite wide application prospect in the field of oil geology. Oil geological drawing vectorization, rock iamge crevice analysis and interwell formation correlation be of quite important significance for oil geological drawing processing, oil prospecting and exploitation. This doctoral thesis describes an up to date research achievement, as well as some existed problems in this field, a series algorithms have been presented and realized based on the practical requirement of engineering application in the following.1. A new color image segmentation algorithm has been proposed by use of fuzzy competitive Hopfield neural network. Mapping image space into gray feature space makes computing speed go up enormously, meanwhile, the optimal performance keeps unchangeableness. Introducing of fuzzy competitive studying resolves the problem that the use of neurons is deficienct. For two class segmentation, it's performance is more superior to Ostu's method. For multi_class segmentation, it has higher processing speed than FCM algorithm.2. According to the characteristic of oil geological drawing, two kinds of algorithms have been proposed in this paper, one is the automatic color separation algorithm based on color histogram analysis, another is the automatic algorithm based on colormap cluster. The former has better robustness; the latter has better real time trait.3. The core of vectorization is to skeleton extraction. The author presents a run_based morphological thinning algorithm and a skeleton extraction algorithm based on feature block. The former can satisfies thinness by use directional structure elements, and the connectedness has been achieved...