Abstract:Image segmentation is a key technique in image analysis for object identification. In this paper, a hybrid segmentation method is proposed, which is based on the Anisotropic Gaussian Kernel (ANGK) edge detector and region adjacent graph (RAG) merging algorithm. An anisotropic directional derivative filter is constructed by angled ANGK to detect the edge contour of original images. Based on the gradient magnitude pattern of the edge contour from ANGK processing, watershed transform is adopted to produce initial partition (coarse segmentation result). Finally, combined with the RAG region merging algorithm, the proposed method performs fine segmentation by merging similar fragmented regions (initial partition) iteratively. Additionally, statistic similarity measure and shape cost function in merging cost are also introduced. They provide quantitative criteria for region merging, which have critical influences on the detection result. A series of experiments are conducted to evaluate the performance of this method, and a preferable localization accuracy as well as noise robustness is proved. Compared with conventional edge and region based methods, the proposed one has a superior segmentation effect. Another advantage is that the problem of oversegmentation can be solved effectively.

Object Contour Extraction Based on Merging Photometric Information with Graph Cuts.

Decompose Image into Meaningful Regions Based on Contour Detector and Watershed Algorithm

Unsupervised Salient Object Segmentation Based on Kernel Density Estimation and Two-Phase Graph Cut

Dynamic Graph Merging For Image Segmentation

A convex neighbor-constrained active contour model for image segmentation

A Hybride Active Contour Model Driven by Global and Local Image Information

Interactive Object Segmentation Using Kernel Density Estimation Based Graph Cuts

Efficiently Solving the Piecewise Constant Mumford-Shah Model using Graph Cuts

Robust Global Minimization of Active Contour Model for Multi-Object Medical Image Segmentation

An Efficient Segmentation Method Based On Dynamic Graph Merging

Precise Cable Contour Extraction Method Based on Candidate Point Merge Search

Iterative Graph Cuts Segmentation with Local Constraints

A Fuzzy Geometric Active Contour Method for Image Segmentation.

Objectness Supervised Merging Algorithm for Color Image Segmentation

Unsupervised Salient Object Segmentation From Color Images

Multi-value Image Segmentation Based on FCM Algorithm and Graph Cut Theory.

A hybrid active contour model based on pre-fitting energy and adaptive functions for fast image segmentation

IMPROVED ACTIVE CONTOUR MODEL AND VISUAL ATTENTION BASED IMAGE SEGMENTATION METHOD

An Active Contour Model Based on Retinex and Pre-Fitting Reflectance for Fast Image Segmentation

A hybrid level set model for image segmentation

Hybrid image segmentation method based on anisotropic Gaussian kernels and adjacent graph region merging