One of the methods for biometric identification is facial features detection, and eye is an important facial feature in the face. In the recent years, automatically detecting eye with different image conditions is att...One of the methods for biometric identification is facial features detection, and eye is an important facial feature in the face. In the recent years, automatically detecting eye with different image conditions is attended. This paper proposes a method which can automatically detect eye in extensive range of images with different conditions. In the proposed method, first an image is enhanced by morphological operations then region of face is detected by hybrid projection function. To identify window of eye, vertical edge dominance map is used. The authors＇ method uses elliptical mask on eye image to detect center of pupil. The mask scans eye image to find minimum gray level because pupil is darkest part in eye image compared with 3 well-known methods. The accuracy of 99.53% on this This method has implemented on JAFFE face database and database confirms efficiency of the proposed method.展开更多
In this paper, we propose a highly automatic approach for 3D photorealistic face reconstruction from a single frontal image. The key point of our work is the implementation of adaptive manifold learning approach. Befo...In this paper, we propose a highly automatic approach for 3D photorealistic face reconstruction from a single frontal image. The key point of our work is the implementation of adaptive manifold learning approach. Beforehand, an active appearance model (AAM) is trained for automatic feature extraction and adaptive locally linear embedding (ALLE) algorithm is utilized to reduce the dimensionality of the 3D database. Then, given an input frontal face image, the corresponding weights between 3D samples and the image are synthesized adaptively according to the AAM selected facial features. Finally, geometry reconstruction is achieved by linear weighted combination of adaptively selected samples. Radial basis function (RBF) is adopted to map facial texture from the frontal image to the reconstructed face geometry. The texture of invisible regions between the face and the ears is interpolated by sampling from the frontal image. This approach has several advantages: (1) Only a single frontal face image is needed for highly automatic face reconstruction; (2) Compared with former works, our reconstruction approach provides higher accuracy; (3) Constraint based RBF texture mapping provides natural appearance for reconstructed face.展开更多
The facial expression recognition systn using the Ariaboost based on the Split Rectangle feature is proposed in this paper. This system provides more various featmes in increasing speed and accuracy than the Haarolike...The facial expression recognition systn using the Ariaboost based on the Split Rectangle feature is proposed in this paper. This system provides more various featmes in increasing speed and accuracy than the Haarolike featrue of Viola, which is commonly used for the Adaboost training algorithm. The Split Rectangle feature uses the nmsk-like shape composed with 2 independent rectangles, instead of using mask-like shape of Haar-like feature, which is composed of 2 --4 adhered rectangles of Viola. Split Rectangle feature has less di- verged operation than the Haar-like feaze. It also requires less oper- ation because the stun of pixels requires ordy two rectangles. Split Rectangle feature provides various and fast features to the Adaboost, which produrces the strong classifier with increased accuracy and speed. In the experiment, the system had 5.92 ms performance speed and 84 %--94 % accuracy by leaming 5 facial expressions, neutral, happiness, sadness, anger and surprise with the use of the Adaboost based on the Split Rectangle feature.展开更多
This paper describes an application where a new KAE （Kansei/Affective Engineering） system was applied to define the properties of the facial images perceived as Iyashi. Iyashi is a Japanese word used to describe a p...This paper describes an application where a new KAE （Kansei/Affective Engineering） system was applied to define the properties of the facial images perceived as Iyashi. Iyashi is a Japanese word used to describe a peculiar phenomenon that is mentally soothing, but is yet to be clearly defined. Instead of analyzing facial expressions of an individual to determine his emotional state, the proposed system introduces a FQHNN （fuzzy-quantized holographic neural network） to find the rules involved in the Kansei evaluation provided by the subjects about the limited dataset of 20 facial images. In order to validate and gain a clear insight into the rules involved in the Kansei evaluation process, Procrustes analysis and CSRBFs （compactly-supported radial basis functions） are combined to generate new facial images. Procrustes analysis is used to find the minimal dissimilarity measure between two facial images with opposite classification （i.e., Iyashi and Non-lyashi）. CSRBFs are proposed for tuning of 17 facial parameters and mapping between facial images within opposite classes. The experiments with two subjects demonstrate that if only two from the five most important parameters of the face are changed, then the Kansei evaluation can change to the opposite class. This paper shows that a continuous and efficient tuning of the design space can be achieved by introducing CSRBF mapping into the new KAE system.展开更多
文摘One of the methods for biometric identification is facial features detection, and eye is an important facial feature in the face. In the recent years, automatically detecting eye with different image conditions is attended. This paper proposes a method which can automatically detect eye in extensive range of images with different conditions. In the proposed method, first an image is enhanced by morphological operations then region of face is detected by hybrid projection function. To identify window of eye, vertical edge dominance map is used. The authors＇ method uses elliptical mask on eye image to detect center of pupil. The mask scans eye image to find minimum gray level because pupil is darkest part in eye image compared with 3 well-known methods. The accuracy of 99.53% on this This method has implemented on JAFFE face database and database confirms efficiency of the proposed method.
基金Project supported by the National Natural Science Foundation of China (Nos. 60533090, 60525108)the National Basic Research Program (973) of China (No. 2002CB312101)+1 种基金the Science and Technology Project of Zhejiang Province, China (Nos. 2005C13032, 2005C11001-05)China-US Million Book Digital Library Project
文摘In this paper, we propose a highly automatic approach for 3D photorealistic face reconstruction from a single frontal image. The key point of our work is the implementation of adaptive manifold learning approach. Beforehand, an active appearance model (AAM) is trained for automatic feature extraction and adaptive locally linear embedding (ALLE) algorithm is utilized to reduce the dimensionality of the 3D database. Then, given an input frontal face image, the corresponding weights between 3D samples and the image are synthesized adaptively according to the AAM selected facial features. Finally, geometry reconstruction is achieved by linear weighted combination of adaptively selected samples. Radial basis function (RBF) is adopted to map facial texture from the frontal image to the reconstructed face geometry. The texture of invisible regions between the face and the ears is interpolated by sampling from the frontal image. This approach has several advantages: (1) Only a single frontal face image is needed for highly automatic face reconstruction; (2) Compared with former works, our reconstruction approach provides higher accuracy; (3) Constraint based RBF texture mapping provides natural appearance for reconstructed face.
基金supported by the Brain Korea 21 Project in2010,the MKE(The Ministry of Knowledge Economy),Koreathe ITRC(Information Technology Research Center)support programsupervised by the NIPA(National ITIndustry Promotion Agency)(NI-PA-2010-(C1090-1021-0010))
文摘The facial expression recognition systn using the Ariaboost based on the Split Rectangle feature is proposed in this paper. This system provides more various featmes in increasing speed and accuracy than the Haarolike featrue of Viola, which is commonly used for the Adaboost training algorithm. The Split Rectangle feature uses the nmsk-like shape composed with 2 independent rectangles, instead of using mask-like shape of Haar-like feature, which is composed of 2 --4 adhered rectangles of Viola. Split Rectangle feature has less di- verged operation than the Haar-like feaze. It also requires less oper- ation because the stun of pixels requires ordy two rectangles. Split Rectangle feature provides various and fast features to the Adaboost, which produrces the strong classifier with increased accuracy and speed. In the experiment, the system had 5.92 ms performance speed and 84 %--94 % accuracy by leaming 5 facial expressions, neutral, happiness, sadness, anger and surprise with the use of the Adaboost based on the Split Rectangle feature.
文摘This paper describes an application where a new KAE （Kansei/Affective Engineering） system was applied to define the properties of the facial images perceived as Iyashi. Iyashi is a Japanese word used to describe a peculiar phenomenon that is mentally soothing, but is yet to be clearly defined. Instead of analyzing facial expressions of an individual to determine his emotional state, the proposed system introduces a FQHNN （fuzzy-quantized holographic neural network） to find the rules involved in the Kansei evaluation provided by the subjects about the limited dataset of 20 facial images. In order to validate and gain a clear insight into the rules involved in the Kansei evaluation process, Procrustes analysis and CSRBFs （compactly-supported radial basis functions） are combined to generate new facial images. Procrustes analysis is used to find the minimal dissimilarity measure between two facial images with opposite classification （i.e., Iyashi and Non-lyashi）. CSRBFs are proposed for tuning of 17 facial parameters and mapping between facial images within opposite classes. The experiments with two subjects demonstrate that if only two from the five most important parameters of the face are changed, then the Kansei evaluation can change to the opposite class. This paper shows that a continuous and efficient tuning of the design space can be achieved by introducing CSRBF mapping into the new KAE system.
