ALFO: Adaptive Light Field Over-Segmentation

Automatic image over-segmentation into superpixels has attracted increasing attention from researchers to apply it as a pre-processing step for several computer vision applications. In 4D Light Field (LF) imaging, aims at achieving not only superpixel compactness and accuracy but also cross-view consistency. Due the high dimensionality of LF images, depth information can be estimated exploited during along with spatial visual appearance features. However, balancing between hybrid features generate robust different images is challenging adequately solved in existing solutions. this paper, an automatic, adaptive, view-consistent method based on normalized cues K-means clustering proposed. Initially, disparity maps all views are entirely improve Afterwards, by using clustering, iteratively divided regular that adhere object boundaries ensure Our proposed automatically adjust weights various characterize each content. Quantitative qualitative results datasets demonstrate outperforming performance terms accuracy, shape regularity view consistency when adaptive weights, compared state-of-the-art methods.