| In this thesis, we present a new algorithm of contrast enhancement of oil sand images. Image contrast enhancement constitutes a crucial pre-processing step in the segmentation of such images, used for granulometry purposes. Oil sand images are difficult to segment due to their complex nature. The new contrast enhancement method aims at improving the quality of the images in order to improve segmentation performance. The method uses a multi-scale image decomposition, obtained with a series of morphological top-hat transformations, where the scale of enhancement corresponds to expected object size. In addition, the new enhancement method is direct, where the level of enhancement is controlled based on a contrast measure. Also, the new method is adaptive, where the enhancement is applied locally, based on local image properties. We present experimental results, where we quantitatively measure the improvement in the quality of oil sand image segmentation, after applying our contrast enhancement method. |
