摘要:
In order to establish connections of plural arithmetic units capable of performing basic functions such as filtering in various connection ways, a bus switch is provided which has a plurality of input data lines connected with output terminals of the arithmetic units, at least one external input data line, a plurality of output data lines connected with input terminals of the arithmetic units, and at least one external output data line. In addition, two register sets are provided which hold arithmetic control information designating contents of processes to be performed by the arithmetic units and connection control information designating connection ways within the bus switch. Depending on the broadcasting system, information held by one of the register sets and information held by the other are updated, and, according to a process algorithm, either one of the two register sets is selected. As a result of such arrangement, a single piece of image processing hardware can be shared between different broadcasting systems as well as between different algorithms.
摘要:
A coprocessor is incorporated in a processor comprising a CPU, an instruction cache, a data memory, a bus controller, an interruption control section and a DMA controller. This coprocessor has a parallel sum-of-products arithmetic operation section, a comparator, an I/O register section, and a sum-of-products factor register section. A frame memory, provided on the input side, stores MUSE or NTSC signals digitized per pixel. The DMA is in control of the transfer of data between the input-side frame memory and the data memory as well as the transfer of data between a frame memory provided on the output side and the data memory. Pixel data stored in the data memory is processed according to broadcasting systems by the switching of sum-of-products factors on the basis of software.
摘要:
In order to establish connections of plural arithmetic units capable of performing basic functions such as filtering in various connection ways, a bus switch is provided which has a plurality of input data lines connected with output terminals of the arithmetic units, at least one external input data line, a plurality of output data lines connected with input terminals of the arithmetic units, and at least one external output data line. In addition, two register sets are provided which hold arithmetic control information designating contents of processes to be performed by the arithmetic units and connection control information designating connection ways within the bus switch. Depending on the broadcasting system, information held by one of the register sets and information held by the other are updated, and, according to a process algorithm, either one of the two register sets is selected. As a result of such arrangement, a single piece of image processing hardware can be shared between different broadcasting systems as well as between different algorithms.
摘要:
A coprocessor is incorporated in a processor comprising a CPU, an instruction cache, a data memory, a bus controller, an interruption control section and a DMA controller. This coprocessor has a parallel sum-of-products arithmetic operation section, a comparator, an I/O register section, and a sum-of-products factor register section. A frame memory, provided on the input side, stores MUSE or NTSC signals digitized per pixel. The DMA is in control of the transfer of data between the input-side frame memory and the data memory as well as the transfer of data between a frame memory provided on the output side and the data memory. Pixel data stored in the data memory is processed according to broadcasting systems by the switching of sum-of-products factors on the basis of software.
摘要:
A plurality of processing elements are connected in cascade so as to constitute a single signal processing apparatus. The signal processing apparatus has a first path for transferring an input data signal and a second path for transferring a processing result of the input data signal. Each of the processing elements has a first input disposed on the first path, a data holding circuit for holding the data signal supplied through the first input, a product-sum circuit for executing a product-sum operation between the data signal held in the data holding circuit and another data signal, a second input disposed on the second path so as to supply the other data signal to the product-sum circuit, a result register for holding the result of the operation calculated by the product-sum circuit and supplying the held result of the operation to the second path, an output selecting circuit for supplying either one of the data signal held in the data holding circuit and the result of the operation held in the result register, and a processing control circuit for controlling respective operations of the data holding circuit, product-sum circuit, result register, and output selecting circuit. Since a route extending from the result register to the output selecting circuit forms a bypass from the second path to the first path, a flexible process can be performed by the signal processing apparatus by using or not using the bypass depending on control information.
摘要:
A plurality of signal processing elements are cascade-connected to form a signal processor having three signal paths. The signal processor is a small-sized device which can be shared by sum-of-products calculation and division. In each signal processing element, first and second shifters and an adder-subtracter are used for performing shift addition for multiplication of a variable by a constant which is a basis of the sum-of-products calculation. The adder-subtracter and a third shifter for shifting a result obtained by the adder-subtracter are used for performing subtraction and shifting for obtaining a partial quotient and a partial remainder of division. The partial quotient thus obtained is transferred to the signal processing element in the next stage through a flag holding circuit.
摘要:
An improved signal processor is disclosed which is suitable for convergence processing of images that achieves parallel processing with a smaller bus structure. An arithmetic array is provided which is formed of ten arithmetic cells capable of concurrently operating. Each arithmetic cell is specified by a denotation of E�(column number),y(row number)! where the column number x is 1.ltoreq.x.ltoreq.4 and the row number y is x.ltoreq.y.ltoreq.4. Each arithmetic cell has a multiplier and an adder for multiply-add operation. An arithmetic cell, specified by E�x,y! where 2.ltoreq.x.ltoreq.4 and x.ltoreq.y.ltoreq.4, receives data from an E�x-1,y! arithmetic cell via a direct bus as well as from an E�x-1,y-1! arithmetic cell via an oblique bus. For example, when pixel data items as to four pixels horizontally arranged in an image are fed to the four arithmetic cells in the first column, the arithmetic cell in the fourth column provides a 4-tap horizontal filter operation result.
摘要翻译:公开了一种改进的信号处理器,其适用于以较小的总线结构实现并行处理的图像的收敛处理。 提供由能够同时操作的十个算术单元形成的运算阵列。 每个算术单元由列号x为1 = x = 4并且行号y为x = y u>的E [(列号),y(行号)]的表示来指定, = 4。 每个算术单元具有用于乘法运算的乘法器和加法器。 由E [x,y]指定的算术单元,其中2 = x 4和x = y 4,通过直接从E [x-1,y]算术单元接收数据 总线以及从E [x-1,y-1]算术单元经由斜线总线。 例如,当在图像中水平排列的四个像素的像素数据项被馈送到第一列中的四个算术单元时,第四列中的算术单元提供4抽头水平滤波器运算结果。
摘要:
This invention discloses an improved signal processor comprising first to third arithmetic units forming a pipeline structure, first to third control information hold circuits each of which holds control information for its corresponding arithmetic unit, first to third selection circuits, and first to third signal transfer circuits. Transfer of a selection signal is delayed by a proportional interval of time to the processing time of each arithmetic unit. In order to perform the switching of arithmetical operations in each arithmetic unit according to the data flow in the pipeline processing, each selection circuit selects among the control information hold circuits depending on the selection signal transferred and provides control information held in a selected control information hold circuit to a corresponding arithmetic unit.
摘要:
According to the signal processing method of the invention, control data is previously generated by using the name of a control signal included in an object signal, and then, the object signal including the control signal is input. The name of the control signal in the control data is substituted with the content of the control signal included in the object signal, and then the object signal is processed by a signal processing unit by using the control data including the content of the control signal. Therefore, the signal processing unit can change the processing to be performed on the object signal in accordance with the content of the control signal included in the object signal.
摘要:
A coprocessor is incorporated in a processor comprising a CPU, an instruction cache, a data memory, a bus controller, an interruption control section and a DMA controller. This coprocessor has a parallel sum-of-products arithmetic operation section, a comparator, an I/O register section, and a sum-of-products factor register section. A frame memory, provided on the input side, stores MUSE or NTSC signals digitized per pixel. The DMA is in control of the transfer of data between the input-side frame memory and the data memory as well as the transfer of data between a frame memory provided on the output side and the data memory. Pixel data stored in the data memory is processed according to broadcasting systems by the switching of sum-of-products factors on the basis of software.