In this paper, we present a new combination of colour and texture informations for image segmentation. This technique is based on principal components analysis of a 3D points cloud, followed by an eigenvalues analysis. A set of colour gradients (morphological, Di-Zenzo) and texture gradients (Gabor, three Haralick attributes, Alternative Sequential Filter (ASF)) are used to test the proposed combination. The segmentation is performed using a hybrid gradient based watershed algorithm. The major contribution of this work consists in combining locally colour and texture informations using an adaptive and non parametric approach. The proposed method is tested on 100 images from the Berkley dataset [1] and evaluated with the Mean Square Error (MSE), the Variation of Information (VI) and the Probabilistic Rand Index (PRI).
展开▼
