Fractal features and fuzzy clustering based tree leaf segmentation from images with the complex background

This paper proposes a new method based on texture and color features to automatically segment the tree leaf from images with the complex background. The tree leaf usually has regular textures while natural background has random textures. Therefore, the local fractal dimension is used as texture features to distinguish the texture difference between the tree leaf and natural background. However, when the background of leaf is rough, the features of the background and the tree leaf might be too close. In this case, the color based fuzzy clustering approach is utilized to segment the leaf image. Experimental results demonstrate the reliability and validity of the proposed algorithm in this paper.