JPS63233480A - Pipeline processing specifying system - Google Patents

Abstract

PURPOSE:To remove necessity for entering a parameter following a command by making a parameter table and a command name as a pair and connecting both the elements through a relational line to specify pipeline processing. CONSTITUTION:A parameter table 1-1 is formed in a memory 1, a parameter necessary for the execution of a command is previously entered and its table number is added to the command. For instance, numerals 1, 2 indicating table numbers entering parameters necessary for execution of through processing are added to a command and the added command THR1+SUB2 is inputted to a processor 2. Input data from a TV camera 7 are set up in a 1st image memory 6-1 and a 1st image processor 4 executes the through processing of the image data and sets up the processed result in a 2nd image memory 6-2. Then, a 2nd image processor 5 executes the subtracting processing between the obtained image data and image data newly set up in the memory 6-1.

Description

[Detailed description of the invention] [Table of contents] Overview Industrial field of application Prior art (Figures 4 and 5) Problems to be solved by the invention Means for solving the problems (Figure 1) Effect Example (Figures 2 and 3) Effects of the invention ($1 required) When performing pipeline processing, the parameters for each process are created in separate tables, and when the user issues a command from the keyboard. , in which the command name and parameter table number are specified as a pair, connected by a relation.

[Industrial application field]

The present invention relates to a pipeline processing setting method, and particularly to a method in which parameters for each process are created in separate tables when performing pipeline processing in which several processes are performed successively.

[Conventional technology]

For example, in the case of transmitting image information, in order to reduce the amount of transmission, research is being conducted on transmitting only the changes compared to the image sampled at the previous time. When performing processing such as extracting changes in an image in this way, pipeline processing is performed using the following command.

THRp i+5tJBp j+ABSpx The first command THR indicates through processing, the second command SUB indicates subtraction processing, the third command ABS indicates absolute value processing,
21% pJ% px each indicates a parameter. The relation + indicates that each command is connected in series.

For example, as shown in FIG. 4, the output image data of the TV left camera 0 is transferred to the first image processing processor 21, second image processing processor 22, third image processing processor 23, first image memory 24, second image memory 25, the third image memory 26, and the fourth image memory 27 to execute each of the above commands, the following processing is performed. It is assumed that the TV left camera 0 outputs an image every 1/30 seconds, and the first image processing processor 21 performs through processing (
The second image processor 22 performs subtraction processing (S
The third image processor 23 performs absolute value processing (AB
S) shall be carried out.

First, image data d1 outputted from the TV left camera 0 at time T1 is passed through processing by the first image processing processor 21 and sent to the first image memory as it is.

At the next time T2, the image data d2 output from the TV left camera 0 is transmitted to the first image processing processor 21 and the second image processing processor 22.

At this time, since the second image processing processor 22 has received the image data d1 at the previous time T1 from the first image memory 24, dl-d2 (subtraction corresponding to each bit and 7 bits) is The result of the subtraction is expressed as the second two's complement.
The positive and negative signs of the subtraction result are set in the third image memory 26. Then, the image data d2 that has been through-processed by the first image processing processor 21 is set in the first image memory 24.

Then, at time T3, in the third image processing processor 23, absolute value processing for d2 set in the second image memory 25 is performed based on the load signal set in the third image memory 26. The result is stored in the fourth image memory 27.

At this time, image data d3 is output from the TV left camera 0, so the second image processing processor 22 performs subtraction processing of d2-d3, and the first image processing processor 21 performs through processing on the image data d3. .

Thereafter, at times T4 and Tc+-;-, each process based on the above-mentioned commands is performed in series in the same manner.

Also, for example, parallel pipeline processing can be performed using the following command.

INVpyXTHRpz The first command INV indicates inversion processing, the second command indicates through processing, and py, , pz each indicate a parameter. The relation x indicates that each command is connected in parallel. In FIG. 5, the first image processing processor 21 inverts the multivalued data output from the TV left camera 0 and transfers it to the first image memory 24.
, and the second image processing processor 22 performs through processing and directly stores the multi-valued data in the second image memory 25.

[Problem that the invention seeks to solve]

When executing the above command, parameters are required. Various parameters necessary for command execution are listed after each command, such as the memory number from which data should be extracted, which address in the memory should be used as the starting address to read image data of a certain size, etc. ing.

In the case of image processing, the amount of parameters is large, so it takes a considerable amount of effort to enter them, and it is easy to make mistakes. In addition, the same parameters may be used again, and each time a long parameter must be entered after the command, making it easy for errors to occur.

An object of the present invention is to provide a pipeline processing specification method that does not require entering parameters after a command.

[Means for solving problems]

In order to achieve the above object, in the present invention, as shown in FIG. 1, a parameter table 1-1 is provided in the memory 1, and parameters necessary for executing the above commands are entered in this parameter table in advance.

Then, a table number containing parameters necessary for executing the command is added to this command.

For example, when performing subtraction processing on an input image after through processing, the following command is input to the processor MPU2.

THR1+5UB2 Here, the number 1.2 indicates that table numbers 1 and 2 are filled with parameters necessary for through processing.

In this case, the processor 2 causes the first image processing processor 4 to execute the command rTHRJ, and causes the second image processing processor 5 to execute the command rsUBJ. TV camera 7
The input data is set to, for example, the first image memory 6-1 via the image input port 3, and the first image processing processor 4 through-processes the image data sent to the first image memory 6-1. and the second image processing processor 5 stores it in the second image memory 6-2.
-2 image data and first image memory 6-1
Subtraction processing will be performed with the newly cent image data.

[Effect]

There is no need to write long parameters after the command name.

〔Example〕

An embodiment of the present invention will be described based on FIGS. 2 and 3.

FIG. 2 is a configuration diagram of an embodiment of the present invention, and FIG. 3 is an explanatory diagram of a parameter table.

In FIG. 2, the same symbols as those in FIG. 1 indicate the same parts.

In the figure, 8 is an image output port, 9 is a third image processor,
10 is a conversion unit, 11 is a terminal device, and 12 is a monitor display unit.

A parameter table 1-1 is prepared in the main memory 1,
As described at 11e in FIG. 3, parameters necessary for executing the command are entered.

The processor 2 controls the entire image processing apparatus as shown in FIG. 2, and performs comprehensive control for creating the parameter table 1-1 and executing input command sequences. .

The image input port 3 is for converting a TV camera and an increment video signal into a no increment signal.

For example, the first image processing processor 4 executes a command r T
It executes HRJ.

The second image processing processor 5 uses the command rSUB, for example.
J.

The image memory 6 is a memory that holds image data output from the TV camera 7 and the results of processing this image data by each image processing processor, and includes a plurality of memories 6-1 to 6-1.
Each memo U 6-1 to 6-4 has a capacity of 1 Mbyte.

The TV camera 7 outputs image data, and although only one is shown in the figure, a plurality of TV cameras are actually used.

The third image processing processor 9 uses, for example, the command rABS.
J.

The conversion unit 10 provides a bus interface between the processor 2 and the local bus.

The terminal device 11 is used to input commands, input parameter table creation data, and display or print out the execution results of command sequences.

The monitor display section 12 performs monitor display for checking the processing status of a command string midway through.

The parameter table 1-1 is a characteristic part of the present invention, and as shown in FIG. 3(a), it contains a proprietary flag FO1 indicating whether the parameter table is in use, an input parameter IP, and parameters related to the image processing processor. PP,
It has parameters OP related to output memory. As many parameter tables as the number of image processing processors to be used are prepared as shown in Figure 3). The input parameter IP in the parameter table indicates the input data to be processed by the image processing processor, and as shown in Figure 3 (C1), the input parameter IP indicates the number of input lights (1 for through processing, subtraction In the case of processing, 2), ID indicating the input light (if processing data from multiple cameras, the camera number is entered; if processing data from multiple memories, the memory number is entered) , block number BLK indicating the memory block position
(For example, memories 6-1 to 6-4 are each 512x5
(divided into four blocks of 12 blocks), the starting point address (X coordinate S
TX, Y coordinates 5TY) and data width (X direction width DX, Y
direction width DY), etc. are entered.

In the parameter PP regarding the image processing processor, the command to be processed by the image processing processor is entered as a numerical code, and in the case of through processing, the command to be processed by the image processing processor is entered in the mask-designated position, for example, opci in FIG. 3(a).
The numeric code r16 of HRJ will be entered.

The parameter OP regarding the output memory specifies the output memory that outputs the results processed by the image processing processor, and includes an output mode MODE indicating the output source, a memory ID indicating the memory number of the cutting destination, and memories 6-1 to 6-1. 6-4
is composed of, for example, eight planes, and the plane select P specifies which one to use.
L.S.L.

BLK and the like specifying the memory block position are written.

The necessary number of such parameter tables are created in advance by the processor 2 by the operator inputting each parameter from the terminal device 11 before inputting a command. Then, each parameter table is managed by adding a number to it. For example, the parameter table for through processing is given table number 1, the parameter table for subtraction processing is given table number 2, the parameter table for absolute value processing is given table number 3, the parameter table for inversion processing is given table number 4, and so on. When performing the processing described with reference to FIG. 4, the following command sequence is input from the terminal device 11 and started.

THR1+5UB2+ABS3 Here, the number next to each command indicates the table number of the parameter table in which parameters necessary for executing the command are entered.

As a result, the processor 2, as shown in FIG. 3(b),
Parameter table (1) with table number 1 necessary for through processing, parameter table (2) with table number 2 necessary for subtraction processing, table number 3 necessary for absolute value processing
For example, the execution of the parameter table Tl) is assigned to the first image processing processor 4, the execution of the parameter table (2) is assigned to the second image processing processor 5, and the execution of the parameter table (Tl) is assigned to the second image processing processor 5. Execution is assigned to the third image processing processor 9.

In Figure 3), Fl is a proprietary flag, the user ID is the operator's name, and the connection code TC is the relation +
is decoded by processor 2 to indicate the relationship between each image processing processor. JCI indicates that the first image processing processor 4 is used in series, and JC2 and JC3 are also This shows that the second image processing processor 5 and the third image processing processor 9 are used in series.

In this way, each image processing processor performs the same processing as in the above case, and the processing results are written into the image memory 6.

Note that parallel processing can also be performed by connecting similarly configured parameter tables in parallel using connection cords.

Of course, the present invention is not limited to the above-mentioned commands, and can be executed in the same manner for other commands. Of course, the number of image processing processors is also the same as 1 above.
It is not limited only to the examples.

〔Effect of the invention〕

According to the present invention, parameters can be obtained by specifying a parameter table number, so pipeline processing with a large number of parameters can be easily and accurately specified in various combinations.

[Brief explanation of drawings]

Figure 1 is a diagram explaining the principle of the present invention, Figure 2 is a configuration diagram of an embodiment of the present invention, Figure 3 is a detailed diagram of a parameter table, Figure 4 is a diagram explaining serial processing in a pipeline, and Figure 5 is a diagram explaining the serial processing in a pipeline. FIG. 2 is an explanatory diagram of parallel processing in a pipeline. ■・-Raw memory l-1・-Parameter table 2-・
Processor 3--Image input port 4--First image processing processor 5-Second image processing processor 6-Image memory 7--TV camera 8--Image output port 9-Third image processing processor

Claims (1)

[Scope of Claims] An image processing apparatus including a plurality of image memories and a plurality of image processing processors includes: a processor (2) that creates a parameter table in which parameters for commands to be processed by the image processing processors are entered; , a holding means (1) for holding a plurality of parameter tables is provided, and when performing pipeline processing in which a plurality of processes are executed in succession, the parameters of each process or the created parameter table and command name are stored in one. A pipeline processing specification method characterized in that pipeline processing is specified by forming pairs and connecting them with relations.