摘要:
An estimation device estimates a hidden state of an estimation subject from an observable state in a manner of a time series. The observable state is observed from the hidden state of the estimation subject under a procedure that has a hierarchical structure, which includes the hidden state of the estimation subject, the observable state, and an intermediate hidden state therebetween. The estimation device includes an estimation subject hidden state predicting means, an intermediate hidden state predicting means based on the state transition structure of the hidden state of the estimation subject, an intermediate hidden state likelihood observing means, an intermediate hidden state estimating means, an estimation subject hidden state likelihood observing means, estimation subject hidden state estimating means, an intermediate hidden state predicting means based on the state transition structure of the intermediate hidden state, and the mixing means.
摘要:
A face model providing portion provides an stored average face model to an estimation portion estimating an affine parameter for obtaining a head pose. An individual face model learning portion obtains a result of tracking feature points by the estimation portion and learns an individual face model. The individual face model learning portion terminates the learning when a free energy of the individual face model is over a free energy of the average face model, and switches a face model provided to the estimation portion from the average face model to the individual face model. While learning the individual face mode, an observation matrix is factorized using a reliability matrix showing reliability of each observation value forming the observation matrix with emphasis on the feature point having higher reliability.
摘要:
A face model providing portion provides an stored average face model to an estimation portion estimating an affine parameter for obtaining a head pose. An individual face model learning portion obtains a result of tracking feature points by the estimation portion and learns an individual face model. The individual face model learning portion terminates the learning when a free energy of the individual face model is over a free energy of the average face model, and switches a face model provided to the estimation portion from the average face model to the individual face model. While learning the individual face mode, an observation matrix is factorized using a reliability matrix showing reliability of each observation value forming the observation matrix with emphasis on the feature point having higher reliability.
摘要:
An estimation device estimates a hidden state of an estimation subject from an observable state in a manner of a time series. The observable state is observed from the hidden state of the estimation subject under a procedure that has a hierarchical structure, which includes the hidden state of the estimation subject, the observable state, and an intermediate hidden state therebetween. The estimation device includes an estimation subject hidden state predicting means, an intermediate hidden state predicting means based on the state transition structure of the hidden state of the estimation subject, an intermediate hidden state likelihood observing means, an intermediate hidden state estimating means, an estimation subject hidden state likelihood observing means, estimation subject hidden state estimating means, an intermediate hidden state predicting means based on the state transition structure of the intermediate hidden state, and the mixing means.
摘要:
The position of a face image within an input image is detected based on results from applying a plurality of weak classifiers in sequence to each of sub-images extracted from the input image. A decision whether to interrupt the sequence and reject a currently extracted sub-image is made based on the sum of a total of weighted decision values obtained up to the current point in the sequence and a total of potential weighted decision values obtainable from the remaining weak classifiers if the extracted sub-image were a face image.
摘要:
The position of a face image within an input image is detected based on results from applying a plurality of weak classifiers in sequence to each of sub-images extracted from the input image. A decision whether to interrupt the sequence and reject a currently extracted sub-image is made based on the sum of a total of weighted decision values obtained up to the current point in the sequence and a total of potential weighted decision values obtainable from the remaining weak classifiers if the extracted sub-image were a face image.
摘要:
A shoe press apparatus of a paper machine for pressing wet paper through a nip includes a counter roll, and a shoe module disposed adjacent to the counter roll. The shoe module includes a cylindrical blanket, and a press mechanism disposed inside the cylindrical blanket, extending in a width direction of the cylindrical blanket and individually pressing a plurality of pressed portions formed on the cylindrical blanket along a traveling direction of the wet paper, toward an outer-surface of the counter roll so that the wet paper is pressed in the nip between the cylindrical blanket and the counter roll so that a pressure can be applied efficiently in response to a density required for paper solely by the shoe press apparatus.
摘要:
A heat insulating/sound absorbing material wherein the entire surface of an inorganic fiber mat is covered with a synthetic resin film or films and a method of producing the same. Front and rear surfaces of an inorganic fiber mat 12 as well as both side surfaces along a longer side thereof are covered with synthetic resin films 13, 14; both ends of the synthetic resin films in the direction of a longer side of the inorganic fiber mat 12 are extended from both ends in the direction of a shorter side thereof to thereby cover both end surfaces in the direction of a shorter side of the inorganic fiber mat 12; an adhesive is applied continuously or discontinuously along the direction of a longer side of film to a portion in the direction of a shorter side of the inorganic fiber mat whereby the inorganic fiber mat and the synthetic resin films, and a mutually opposing portion of the synthetic resin films are bonded so as to make air communicate through unbonded portions.
摘要:
An improved multi-layer paper sheet forming system is disclosed, in which a degree of dispersion of fibers under the condition where a paper layer has been formed is large, thereby formation of a paper sheet is improved, moreover, bases of short wire cloth sections can be installed at the same level as a building floor and a complicated base of the shape of steps is not necessitated. The system comprises one endless felt loop and a plurality of short wire cloth sections disposed on the endless felt and provided with a dewatering instrument having, as dewatering elements, forming blades consisting of two zones, that is, a flat front half and a curved rear half. The short wire cloth section includes a forming roll having a large diameter and a breast roll having a small diameter, and the plurality of short wire cloth sections are disposed substantially at the same level. A movable return roll is equipped, which raises the felt between the adjacent short wire cloth sections so as to wrap around the short wire cloth so that an initial engagement point between the felt and the short wire cloth may come to an arbitrary position of the curved portion of the dewatering instrument existing at the upper portion of the short wire cloth section.
摘要:
A dewatering instrument of a twin-wire former allows the angles at which the wires are wrapped around the dewatering blades to be adjusted even during operation so that pulsating pressure applied to the paper stock may be set appropriately according to the prevailing paper making condition. The respective dewatering blades of the dewatering instrument are supported by two support bodies, the first one of which is fixed and pivotably supports the blade, and the second one of which is movable to pivot the blade about the first. For example, the second support body may be a flexible tube which is inflatable and deflatable. When the blade is pivoted, the attitude of the land of the blade is changed so that the wrap angles of the wires with respect to the land of the dewatering blade are adjusted.