Automatic detection of frontal face midline by chain-coded Merlin-Farber Hough trasform

Daichi Okamoio, Wataru Ohyama, Tetsushi Wakabayashi, Fumitaka Kimura

Research output: Contribution to journalArticlepeer-review

2 Citations (Scopus)


We propose a novel approach for detection of the facial midline (facial symmetry axis) from a frontal face image. The facial midline has several applications, for instance reducing computational cost required for facial feature extraction (FFE) and postoperative assessment for cosmetic or dental surgery. The proposed method detects the facial midline of a frontal face from an edge image as the symmetry axis using the Merlin-Faber Hough transformation. And a new performance improvement scheme for midline detection by MFHT is present. The main concept of the proposed scheme is suppression of redundant vote on the Hough parameter space by introducing chain code representation for the binary edge image. Experimental results on the image dataset containing 2409 images from FERET database indicate that the proposed algorithm can improve the accuracy of midline detection from 89.9% to 95.1 % for face images with different scales and rotation.

Original languageEnglish
Pages (from-to)2159-2166+9
JournalIEEJ Transactions on Electronics, Information and Systems
Issue number12
Publication statusPublished - 2010

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering


Dive into the research topics of 'Automatic detection of frontal face midline by chain-coded Merlin-Farber Hough trasform'. Together they form a unique fingerprint.

Cite this