AWkS: adaptive, weighted k-means-based superpixels for improved saliency detection

Abstract

Clustering inspired superpixel algorithms perform a restricted partitioning of an image, where each visually coherent region containing perceptually similar pixels serves as a primitive in subsequent processing stages. Simple linear iterative clustering (SLIC) has emerged as a standard superpixel generation tool due to its exceptional performance in terms of segmentation accuracy and speed. However, SLIC applies a manually adjusted distance measure for dis-similarity computation which directly affects the quality of superpixels. In this work, self-adjustable distance measures are adapted from the weighted k-means clustering (W-k-means) for generating superpixel segmentation. In the proposed distance measures, an adaptive weight associated with each variable reflects its relevance in the clustering process. Intuitively, the variable weights correspond to the normalization terms in SLIC that affect the trade-off between superpixels boundary adherence and compactness. Weights that influence consistency in superpixel generation are automatically updated. The variable weights update is accomplished during optimization with a closed-form solution based on the current image partition. The proposed adaptive, W-k-means-based superpixels (AWkS) experimented on three benchmarks under different distance measure outperform the conventional SLIC algorithm with respect to various boundary adherence metrics. Finally, the effectiveness of the AWkS over SLIC is demonstrated for saliency detection.

Reversible data hiding algorithm for color images combining superpixel segmentation and pixel value ordering

“…In the literature, only a few researchers have applied superpixels to the field of information hiding. In this paper, a superpixel segmentation method of simple linear iterative clustering 26 is adopted; it segments the image into super pixel blocks with irregular shapes and high color similarity and takes advantage of the high color similarity of the super pixel blocks, which greatly improves the number of smooth blocks and the prediction accuracy of the PVO predictor, reduces the invalid translation, and improves the quality of the steganographic image. Different PVO embedding methods are used for the superpixel blocks in the R plane and the B plane to maximize the consistency of the embedding positions in the planes to reduce the modification of the G plane when adjusting the gray value and further improve the quality of the steganographic image.…”

Salient Object Detection by LTP Texture Characterization on Opposing Color Pairs under SLICO Superpixel Constraint

“…The color micro-textures' boundaries correspond to the superpixel obtained thanks to the SLICO (Simple Linear Iterative Clustering with zero parameter) algorithm [33] which is faster and exhibits state-of-the-art boundary adherence. We would like to point out that there are other superpixels algorithms that have a good performance such as the AWkS algorithm [34]; however, we chose SLICO because it is fast and almost parameter-free. A feature vector representing the color micro-texture is obtained by the concatenation of the histograms of the superpixel (defining the micro-texture) of each opposing color pair.…”