Learning to Detect Boundaries in Natural Image Using Texture Cues and EM

Most unsupervised methods in boundary detection fail to manage the small veins with strong contrast in brightness. Aiming at this, the paper presents a novel method in boundary detection, which is based on two parts. The first part is combination of LBP (local binary pattern) and maximum difference...

Full description

Saved in:
Bibliographic Details
Main Authors: Yan Li, Siwei Luo, Qi Zou
Format: Conference Proceeding
Language:English
Subjects:
boundary detection
Brightness
Colored noise
Computer science
Computer vision
Conference management
Detectors
Image edge detection
Image segmentation
Performance gain
texture
Veins
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Most unsupervised methods in boundary detection fail to manage the small veins with strong contrast in brightness. Aiming at this, the paper presents a novel method in boundary detection, which is based on two parts. The first part is combination of LBP (local binary pattern) and maximum difference criterion of texture to get a clear salient-boundary-point image, using local texture cues to cut down the insignificant edges. In the second part we use a new EM framework including salient cue to approximate the points. We choose The Berkeley Segmentation Dataset and Benchmark as our estimate criterion. Experimental results show the model gain good performance on extracting the object boundary.
ISSN:2157-9555
DOI:10.1109/ICNC.2008.233