Abstract                         Publications                  Results                  Back to my Homepage

Back to top

There are several kinds of textures caused by the albitic phase inside perthite crystals (one of the minerals found in granite rocks). Computerized classification of perthite images can help geologists distinguish between granitoid rocks and contribute to the understanding of the influence of various physical conditions on the development of the different perthite textures. Today, the classification is done by hand, a procedure that demands very much work and it’s results can not be a basis for a large scale research since they are strongly dependent on the person doing the classification. The system presented in this work has three main parts. At the first part the borders of the crystal of interest are found using color and texture features. At the second part the color image is transformed into a binary one and noises are cleaned. At the third part connected albite elements are disconnected using an operator that finds vein elements and an iterative operation repetition that disconnects connected patch elements. After separating the elements, classification is done using the element dimensions. The system results were compared with hand made border location and classification results using a set of perthite crystals containing elements from all texture classes. These results show that in most cases the border location fits very well the true crystal border and there is a very high compatibility between the system classification results and the manual ones. The main contribution of the proposed system is that it is a fast and objective tool for perthite crystal texture classification and will enable a large scale research of the perthite texture aspect of granite rocks.

Back to top

Publications-

Back to top

Back to top

 
Back to my Homepage