TY - GEN
T1 - Face sketch multiple features detection using simultaneously shape and landmark movement
AU - Hariadi, Mochamad
AU - Muntasa, Arif
AU - Purnomo, Mauridhi Hery
PY - 2009
Y1 - 2009
N2 - Nowadays, retrieving a person identity using a photograph from the face image database is a crucial job especially in police investigations. Unfortunately in many cases,the photo image of a suspect is not available. Only a face sketch drawing based on the recollection of an eyewitness is available. Usually, there are two kind of face sketches employed in police investigations i.e. halftone face sketches. In this paper, we propose a modified line gradient method called Maximum Line Gradient Method to detect multiple features from halftone face sketches by using simultaneously moving shapes and landmarks. Our proposed method is divided into four stages: training, create image gradient, shape initialization, and multiple features detection processes. The last stage is started by searching the maximum line gradient value between two landmarks. Thus, by using the Similarity Transformation Equation, the set of landmarks (shape) will be simultaneously moved. The position of new landmark is enhanced by using simultaneously landmark movements on each shape. In the experiment, we employ 50 halftone face sketches which being examined by using 7 features with 38 landmarks. Our propose method demonstrates that the detection accuracy is 92.16%.
AB - Nowadays, retrieving a person identity using a photograph from the face image database is a crucial job especially in police investigations. Unfortunately in many cases,the photo image of a suspect is not available. Only a face sketch drawing based on the recollection of an eyewitness is available. Usually, there are two kind of face sketches employed in police investigations i.e. halftone face sketches. In this paper, we propose a modified line gradient method called Maximum Line Gradient Method to detect multiple features from halftone face sketches by using simultaneously moving shapes and landmarks. Our proposed method is divided into four stages: training, create image gradient, shape initialization, and multiple features detection processes. The last stage is started by searching the maximum line gradient value between two landmarks. Thus, by using the Similarity Transformation Equation, the set of landmarks (shape) will be simultaneously moved. The position of new landmark is enhanced by using simultaneously landmark movements on each shape. In the experiment, we employ 50 halftone face sketches which being examined by using 7 features with 38 landmarks. Our propose method demonstrates that the detection accuracy is 92.16%.
KW - Active shape model
KW - Face sketch
KW - Feature detection
KW - Line gradient
KW - Police investigations
UR - http://www.scopus.com/inward/record.url?scp=77649295557&partnerID=8YFLogxK
U2 - 10.1109/SoCPaR.2009.81
DO - 10.1109/SoCPaR.2009.81
M3 - Conference contribution
AN - SCOPUS:77649295557
SN - 9780769538792
T3 - SoCPaR 2009 - Soft Computing and Pattern Recognition
SP - 381
EP - 386
BT - SoCPaR 2009 - Soft Computing and Pattern Recognition
T2 - International Conference on Soft Computing and Pattern Recognition, SoCPaR 2009
Y2 - 4 December 2009 through 7 December 2009
ER -