Texture segmentation using vector-valued Chan–Vese model driven by local histogram

• The local histogram is introduced into the vector-valued Chan–Vese model for texture segmentation.
• A Bayesian method is introduced to determine the optimal number of bins in the local histogram.
• The local binary pattern is employed to segment textures of inhomogeneity.
• The effectiveness of the proposed approach is demonstrated by various examples.

A novel region-based active contour is proposed for texture segmentation. The proposed method is based on the vector-valued Chan–Vese model and local histogram, and the Wasserstein distance is employed to measure the distance between two histograms. Since the histogram is a powerful tool to characterize texture, the proposed method behaves effectively to segment different texture region. Moreover, a Bayesian method is adopted to determine an optimal number of bins in the histogram, so that the computation load can be reduced considerably whilst the effectiveness of histogram to represent texture remains unchanged. Experiments and comparison are conducted and the results show that the proposed strategy is effective for texture segmentation.

Figure optionsDownload as PowerPoint slide