Abstract:We investigate a fundamental problem in computer vision: unsupervised image segmentation. During the last decade, the Normalized Cuts has become very popular for image segmentation. NCuts guarantees a globally optimal solution in the continuous solution space, however, how to automatically select the number of segments for a given image is left as an open problem. Recently, the lossy minimum description length (LMDL) criterion has been proposed for segmentation of images. This criterion can adaptively determine the number of segments, however, as the optimization is combinatorial, only a suboptimal solution can be achieved by a greedy algorithm. The complementarity of both criteria motivates us to combine NCuts and LMDL into a unified fashion, to achieve a better segmentation: given the NCuts segmentations under different numbers of segments, we choose the optimal segmentation to be the one that; minimizes the overall coding length, subject to a given distortion. We then develop a new way to use the coding length decrement as the similarity measure for NCuts, so that our algorithm is able to seek both the optimal NCuts solution under fixed number of segments, and the optimal LMDL solution among different numbers of segments. Extensive experiments demonstrate the effectiveness of our algorithm.

Unsupervised Segmentation of Natural Images Via Lossy Data Compression

Rough-fuzzy set approach for color and texture based image segmentation

Multispectral image compression by cluster-adaptive subspace representation

Segmentation of Natural Images by Texture and Boundary Compression

Segmentation of Multivariate Mixed Data Via Lossy Coding and Compression

Natural image segmentation with adaptive texture and boundary encoding

Texture Segmentation Using Kernel-based Techniques

Learnable Pixel Clustering Via Structure and Semantic Dual Constraints for Unsupervised Image Segmentation.

Multiresolution Segmentation of Video Objects in the Compression Domain

Image Segmentation For Improved Lossless Screen Content Compression

Unsupervised Image Segmentation Based On Contourlet Texture Features And Byy Harmony Learning Of T-Mixtures

A Dynamically Weighted Loss Function for Unsupervised Image Segmentation

Towards Hypothesis Testing and Lossy Minimum Description Length: A Unified Segmentation Framework

Segmentation Via Ncuts And Lossy Minimum Description Length: A Unified Approach

Unsupervised Texture Segmentation Via Applying Geodesic Active Regions To Gaborian Feature Space

Color Image Segmentation Using Density-Based Clustering

Image Segmentation Based on Non-convex Low Rank Multiple Kernel Clustering

Residual-driven Fuzzy C-Means Clustering for Image Segmentation

Unsupervised image segmentation based on local pixel clustering and low-level region merging

Comparison of Segmentation-Based Image Compression Using Threshold, Region Growing, and Edge Detection

Image-driven Unsupervised 3D Model Co-Segmentation