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
Storage management strategy can be expressed by a file system. Commercial file system for embedded application is generally complicated and resource wasted. In this paper, a specified file system adapted to embedded s...Storage management strategy can be expressed by a file system. Commercial file system for embedded application is generally complicated and resource wasted. In this paper, a specified file system adapted to embedded system with flash-based memory is developed. To guarantee the average usage of flash storage sectors, the strategy of wear leveling and adaptive damage management is introduced, in which a dynamic storage space management mode and the strategy of first in first out (FIFO) are adopted. Moreover, the strategy of redundancy design and fast-calculation and tracing is also adopted to extend the life of kernel sector, which can guarantee the reliable service of system booting. The practical application in an embedded CNC (computerized numerical control) platform proves that the file system has effective performance. Furthermore, the flash file system can be transplanted to different embedded platforms by changing a few bottom hardware parameters with universality.展开更多
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 National Natural Science Foundation of China (No50475117)Specialized Research Fund for the Doctoral Program of Higher Education of China (No20060056016)
文摘Storage management strategy can be expressed by a file system. Commercial file system for embedded application is generally complicated and resource wasted. In this paper, a specified file system adapted to embedded system with flash-based memory is developed. To guarantee the average usage of flash storage sectors, the strategy of wear leveling and adaptive damage management is introduced, in which a dynamic storage space management mode and the strategy of first in first out (FIFO) are adopted. Moreover, the strategy of redundancy design and fast-calculation and tracing is also adopted to extend the life of kernel sector, which can guarantee the reliable service of system booting. The practical application in an embedded CNC (computerized numerical control) platform proves that the file system has effective performance. Furthermore, the flash file system can be transplanted to different embedded platforms by changing a few bottom hardware parameters with universality.
