This paper presents a novel variational model based on fuzzy clustering and total variation regularization for superpixel segmentation. Compared with the classical hard-labeled methodologies, our approach gives soft results via the fuzzy membership function, and moreover, the use of total variation provides additional information that can enhance the superpixel regularity, which in turn improves the segmentation performance. To efficiently minimize the energy functional of the proposed model, we adopt an alternating direction method of multipliers with the modified Chambolle's fast duality projection algorithm. Our algorithm can generate regular and compact superpixels with high segmentation accuracy, satisfactory boundary adherence and low computational cost. Comparative experimental results with the current state-of-the-art approaches reveal the superior performance of the proposed method.
This paper presents a novel variational model based on fuzzy region competition and statistical image variation modeling for image segmentation. In the energy functional of the proposed model, each region is characterized by the pixel-level color feature and region-level spatial/frequency information extracted from various image domains, which are modeled by the windowed bit-plane-dependence probability models. To efficiently minimize the energy functional, we apply an alternating minimization procedure with the use of Chambolle’s fast duality projection algorithm, where the closed-form solutions of the energy functional are obtained. Our method gives soft segmentation result via the fuzzy membership function, and moreover, the use of multi-domain statistical region characterization provides additional information that can enhance the segmentation accuracy. Experimental results indicate that the proposed method has a superior performance and outperforms the current state-of-the-art superpixel-based and deep-learning-based approaches.
This paper presents a superpixel segmentation algorithm that integrates anisotropic total variation regularization within a fuzzy clustering framework. While isotropic total variation is well-known for its edge-preserving properties, its non-adaptive nature often leads to over-regularization. In contrast, the anisotropic model formulates superpixel regularity in relation to image contours, thereby preventing the loss of image details in areas of high contour density during optimization. Compared to classical segmentation algorithms that employ non-adaptive regularization, the proposed content-adaptive approach enhances superpixel regularity while maintaining boundary adherence to image contours. Furthermore, to optimize the functional effectively, an alternating direction method of multipliers along with the enhanced Chambolle’s fast duality projection algorithm are employed. Competitive experiments against existing regular segmentation algorithms demonstrate that our proposed methodology achieves superior performance in terms of boundary recall, compactness, and shape regularity criteria, outperforming these methods by an average of at least 3%, 5%, and 3%, respectively. Furthermore, when compared with irregular segmentation algorithms, our approach achieves the best results in terms of compactness, contour density, and shape regularity criteria, with average improvements of at least 56%, 22%, and 45%, respectively.