摘要:
The method allows the creation of 3-D facial models, which can be used, for instance, for the avatar implementation, video-communication applications, video games, video productions, and for the creation of advanced man-machine interfaces. At least one image of a human face is provided together with a 3D facial model (M) having a vertex structure and comprising a number of surfaces chosen within the set formed by a face surface (V), surfaces of the right eye (OD) and left eye (OS), respectively, and surfaces of the upper teeth (DS) and lower teeth (DI), respectively. Among the vertices of the structure of the model (M) and on such at least one face image, respective sets of homologous points are chosen. The model structure (M) is then modified in such a way that the above respective sets of homologous points are made to coincide.
摘要:
A charge pump (2) is supplied to reset in rated conditions the error signal of a phase-locked loop of the type whereby a phase detector (1) periodically supplies this pump (2) with a first and second impulse having emission instants dependent on the phase ratio between phase-locked loop input signals and are allocated to control circuit output increase or decrease respectively by means of ring filter (3a, 3b). The pump features circuits (21, 22, 23, 24, 28) to transform the first and second impulse into a first and second voltage signal of longer duration than maximum impulse duration, the voltage signals having values the difference of which depends on the phase ratio between the loop input signals, and to generate a current signal representative of such difference.
摘要:
A receiver is described, for instance for GSM systems, wherein a plurality of digital signals each comprising a training sequence (x p (t)) usable to generate an estimation of the transmission channel are received in diversity (A.1 ... A.N) in the form of a plurality of replicas (10.1 ... 10.N) each comprising a respective replica of the training sequence. The received digital signal is subjected to a delaying action thereby obtaining a plurality of versions of such signal, each comprising a respective set of signal replicas. Each version is subjected to a respective filtering action (23.1 ... 23.N) independently from the other versions, by employing a respective first set of filtering coefficients (w* 11 ... w* N1 ... w* Nj ) obtained starting from a respective initial set of filtering coefficients. This respective initial set is obtained by performing, for each replica of the received signal, a filtering action exploiting a respective second set of filtering coefficients identified as a signal (u (t)) able to generate, by convolution (u (t) * x p (t)) with the training sequence, a unitary function δ (t) on a given time window, preferably on one bit. The filtering coefficients of the first set are derived starting from said initial set of filtering coefficients through an MMSE adaptation, preferably with a initial rapid convergence phase of the RLS type followed by an optimisation phase of the LMS type.
摘要:
A memory for searching information through prefix analysis, in particular for building routing tables for nodes of high speed communication networks, such as Internet network, comprises a memory element (M1) which stores a set of information items associated each one to a mask information indicative of the number of significant characters in the respective prefix and to a target information. For the implementation of a search criterion based on the longest prefix match, each cell comprises an information field that provides either an address of a next row for the continuation of a search or an information relating to a target reached, and a pair of flags (GO, TARGET) specifying the contents of the information field. An auxiliary vector (AUX), which comprises as many cells as many the memory rows are, is arranged to store, when the flags in a cell in the memory element indicate the reaching of a target together with the need of prosecuting search operations in a next row, the target information in its cell associated to said next row.
摘要翻译:一种通过前缀分析用于构建路由表用于高速通信网络,:诸如因特网的网络的节点搜索信息,在特定的存储器,包括:存储器元件(M1),其存储相关联的一组信息项每一个以表示一个掩模信息 的在respectivement前缀和目标信息显著字符数。 对于搜索条件的基础上,最长前缀匹配的实现中,每个单元包括(信息字段并提供下一行的地址的任一搜索或信息与目标达成的延续,以及一对标志 GO TARGET)指定的信息字段的内容。 辅助载体(AUX),其包括许多存储器行,被安排为存储尽可能多的细胞,当在存储元件单元中的标志指示一个目标的连同有需要的起诉搜索操作到达 下一行,关联到所述下一行其小区的目标信息。
摘要:
A current reference generator comprises a pair of identical units (G1, G2) which generate respective current references (I1, I2), and a circuit (CL) for the linear combination of the two references. In each of the two units, the elements (S1, S2) which, by their current-voltage characteristics, determine the working point comprise respectively a single transistor (T1) and a pair of transistors (T2, T3), of the same type as the first, connected in series. A differential amplifier (AD) maintains stable the working point of the respective unit as power supply voltage varies.
摘要:
A laser module (1) comprises a Fabry-Perot cavity active element (2) with a facet (3) bearing an anti-reflection coating and an external cavity made by an optical fibre Bragg grating (5) with low reflectivity. This grating (5) presents a non-uniform profile of modulation of the refractive index, asymmetrical in the direction of the grating length, such as to give rise to a position of the equivalent mirror plane that is offset towards the end of the grating (5) that is closer to the active element (2).
摘要:
A grating (10) made in the core of an optical fibre (1) presents a non-uniform and asymmetrical profile of the modulation of the refractive index in the direction of the length, which profile is represented by a curve that rises gradually and monotonically from a minimum and substantially null value, with substantially horizontal tangent, in correspondence with an end of the grating (10), until a maximum value, also with substantially horizontal tangent, which is reached in correspondence with the other end of the grating (10).
摘要:
An active single mode optical fiber has the core made of a rare earth doped non-oxide glass and the cladding made of an oxide glass. The glass of the core has a melting temperature lower than that of the glass of the cladding and lying within the range of the softening temperatures of the latter. In a preferred embodiment the core is made of a chalcogenide glass and the cladding is made of a lead silicate glass. To produce the fibre, a preform, obtained by introducing an element made of the non-oxide glass into the hole (2) of a capillary tube (1) made of the oxide glass, is brought to a temperature lying within the range of softening temperatures of the oxide glass and not lower than the melting temperature of the non-oxide glass, and is drawn. The capillary tube (1), during the drawing process, serves as a container for the molten glass of the core.
摘要:
An integrated circuit for manipulating digitised video sequences is provided, for use in a system for transmission and reception of compressed video sequences to perform, possibly with the aid of an external memory, re-ordering, format conversion, prediction and motion compensation on the pictures in a sequence. The device (MSM) comprises means (MT1, MT2) for temporarily storing sequences to be manipulated and data read from the external memory; means (DIB) for decoding information about the manipulations to be performed; addressing means (ISS, CM, CME) for transferring the data between the device and the external memory; means (IU, RC) for configuring the device by means of a remote processing unit; means (IHV) for processing the data read from the external memory; means (SU) for arranging the output sequences in the format required by the function to be performed. A controller may control, supervise and set up the functions to be performed.