JPH03103986A - Image processing system - Google Patents

Abstract

PURPOSE:To extract the feature quantity to a frame immediately after the change of a parameter without increasing a load of a software, and to operate the system at a high speed by bringing a module for processing each image data of continuous inputs, and a module for extracting the feature quantity from the processed image data to pipeline coupling. CONSTITUTION:In a module 12 for extracting the feature quantity, the number of image pieces corresponding to the waiting time determined from the number of pieces counted by a module 10 for processing an image and the number of pieces counted by the module 12 for extracting the feature quantity is set by a setting means 18. In this case, a change instruction of a parameter is stored by a storage means 20 in advance, and when the number of pieces of processing images inputted from the time of setting the number of image pieces reaches the set number of pieces, it is decided that the waiting time is ended and an instruction is executed, and in accordance with the instruction, the parameter for extracting the feature quantity is changed. In such a way, even against the frame immediately after the parameter is changed, the feature quantity of an exact value can be extracted, and the system can be operated at a high speed.

Description

[Detailed Description of the Invention] [Table of Contents] Overview Industrial Field of Application Conventional Technology Problems to be Solved by the Invention Means for Solving the Problems Actions Examples Effects of the Invention [I! [Required] Perform processing such as binarization and contour extraction on image data,
Regarding the image processing system that extracts feature quantities such as density histograms and projection values from the image data, it is possible to extract accurate values of the feature quantities for the frame immediately after parameter changes without increasing the software load, increasing the speed of the system. With the aim of providing an image processing system that can realize image processing, a module for processing continuous input image data and a module for extracting features from the processed image data are combined in a pipeline to create an image processing module. includes a means for counting the number of input image data from the vertical synchronization signal, and a module for feature extraction includes means for counting the number of input images from the vertical synchronization signal, and a module for image processing. The number of sheets counted and inside.・Means for setting the number of images corresponding to a fixed waiting time based on the counted number of images, means for storing a parameter change command, and when the number of processed images input from the time of setting the number of images reaches the set number. , according to the memorized instructions. and means for changing parameters for feature extraction.

[Industrial Application Field] The present invention is applied to an image processing system that performs processing such as binarization and contour extraction on image data and extracts feature quantities such as density histograms and projection values from the image data. It is.

In recent years, when automating product inspection and product classification using image processing technology, image processing systems that can perform high-speed processing such as detecting the target object by binarizing it and then measuring the area and perimeter using histogram calculations are becoming increasingly popular. requested.

[Prior Art] A conventional image processing system of this type is constructed as shown in FIG.

As can be understood from the figure, an image data processing module 30 that performs binarization, contour extraction, etc., and a feature amount extraction module 32 that performs feature amount extraction such as histogram calculation are coupled in a pipeline. There may be a plurality of image processing modules 30).

Here, synchronization signals corresponding to vertical and horizontal signals are input to each module 30, 32 together with image data, and are output after a predetermined vibe line delay time τ determined for each module 30, 32 has elapsed.

A time chart at this time is shown in FIG.

Parameters of each module 30 and 32 are changed in synchronization with the beginning of the vertical synchronization signal.

Here, if the parameters of the image processing module 30 are changed at time tl, and a feature extraction command is issued to the feature extraction module 32 at time t2 immediately thereafter, the module 32 changes the parameters of the module 30. This means that feature amount extraction will be performed on the image data before the extraction.

Therefore, in order to extract accurate values (values after parameter changes), it is necessary to delay the timing of issuing commands to the module 32.

This delay time is the sum of the time T from the time tl at which parameters are set for the module 30 to the beginning of the next frame (frame 2 in this case) and the vibration line delay time τ.

However, this type of system does not have a mechanism to notify the host side of the above-mentioned time T, so 33 msec, which is the worst value of the delay time, is regarded as the time T, and this is set in software to avoid waiting time. First, we had to decide when to issue the command.

For this reason, in the conventional method, a time loss occurs every time a parameter is changed, and it is not possible to extract a feature with an accurate value for the frame immediately after changing the parameter (frame 2' in Figure 8). There was a drawback.

[Problems to be Solved by the Invention The present invention has been made in view of the above-mentioned conventional drawbacks,
The purpose is to provide an image processing system that can extract accurate values of feature amounts for frames immediately after parameter changes without increasing the software load, and that can speed up the system.

[Means for Solving the Problems] In order to achieve the above object, an image processing system according to the present invention is configured as shown in FIG.

In the figure, a module 10 that processes each image data that is continuously input and a module 12 that extracts feature amounts from the processed image data are coupled in a pipeline.

The image processing module 10 includes means 14, and the number of input image data is counted by the means 14 from the vertical synchronization signal.

Further, the feature extraction module 12 includes a counting means 16.
, a setting means 18, a storage means 20, and a changing means 22, and the number of input images is counted by the counting means 16 from the vertical synchronization signal.

In addition, the number of images counted by the image processing module 10 and the feature quantity extraction module 12 are determined by the setting means 1.
The number of images corresponding to the waiting time is set based on the number of images counted.

The storage means 20 stores a parameter change command, and the change means 22 extracts the feature amount according to the stored command when the number of processed images input from the time of setting the number of images reaches the set number. parameters are changed.

[In the module 12 for feature extraction, the number of images corresponding to the waiting time is determined from the number counted by the image processing module 10 and the number counted by the feature extraction module 12, and is set by the setting means 18. Set by

Here, the command to change the parameter is stored in advance in the storage means 20, and when the number of processed images input from the time of setting the number of images reaches the set number, the command is executed as if the waiting time has ended, and the command is executed. Parameters for feature extraction are changed according to the command.

Therefore, feature quantity extraction with accurate values can be performed even for the frame immediately after the parameter has been changed.

[Embodiment] Hereinafter, a preferred embodiment of the image processing system according to the present invention will be described based on the drawings.

FIG. 2 shows an explanation of the overall configuration of the embodiment, which includes a plurality of modules 10-1.1 for processing image data. 1
0-2. . . 10-n and the feature extraction module 12 are connected in a pipeline, and each module 10-1. 10-2. . . 10-n, 12 are connected to the host 42.

The schematic configuration of the module 1o that processes image data is as follows.
The counter 40 includes a counter 40 for counting vertical synchronization signals and an image processing section 41, and the counter 4
A value of 0 is configured to be readable by the host 42.

The schematic configuration of the module 12 for extracting feature quantities is shown in FIG. is configured so that it can be read from the host 42.

Further, the module 12 for extracting features includes a reset signal generation circuit 46, a down counter 48, and a zero determination circuit 50.
, a write signal generation circuit 52, an instruction register 54, a control register 56, and a feature extractor 58.

Here, the operation of this embodiment will be explained.

First, when starting the image processing system, each module 10.1
The second vertical synchronization signal counting counters 40 and 44 are reset.

Immediately after the parameters are set in the image data processing module 1o, the value (FC,a) of the counter 4o of the image processing module 10 is read.

Next, the counter 4 of the feature extraction module 12 connected to the vibrator after the image processing module 10
A value of 4 (Feb) is read.

10- Here, a value (set number of sheets) corresponding to the waiting time is set in the down counter 48 of the feature quantity extraction module 12 from the values of both counters 40 and 44.

Its value is derived from the following formula.

Waiting time value (set number of sheets) = FCa-Feb+1 Here, the down counter 48 uses the set value (set number of sheets) as the initial value of the waiting time, and then the value of the down counter 48 counts down according to the vertical synchronization signal. be done.

Then, when the value of the down counter 48 becomes zero,
The output is input to the zero judgment circuit 50, and the zero judgment circuit 5
0 is regarded as the end of the waiting time and causes the write signal generation circuit 52 to generate a write signal.

When the write signal is input to the control register 56, the command written in the instruction register 54 is output to the feature amount calculation section 58, and thereby the feature amount extraction section 58 extracts the feature amount.

Here, when a write signal is input to the control register 56,
The reset signal generating circuit 46 generates a reset signal, and the reset signal is input to the down counter 48, so that the down counter 48 is placed in a reset state.

FIG. 6 shows a time chart in the case where the image processing parameters are issued at time tl and the feature quantity extraction command is issued at time t2.

As can be understood from the figure, at the same time as the parameter change is set at time ti, a value corresponding to the waiting time (for example, two frames) is set on the feature extraction module 12 side.

A set value corresponding to this waiting time is given to the down counter 48 based on the value of the counter 40.44.

Then, the instruction is written to the instruction register 54 at time t2.

Here, the feature quantity extraction module 12 side is in a standby state for the set waiting time, and the frame 2 shown in FIG.
When ' is input, the waiting time is over, and a command is issued from the control register 56 to the feature extracting section 58, thereby starting extraction of the feature.

12-Thus, according to the system of this embodiment, feature amounts can be extracted for a frame immediately after a parameter change, and the speed of the system can be increased.

FIG. 8 is a time chart explaining the operation of the conventional example; There is.

[Effects of the Invention] As described above, according to the present invention, feature amounts can be extracted for a frame immediately after a parameter change without increasing the software load, and system speedup can be achieved.

[Brief explanation of drawings]

Fig. 1 is an explanatory diagram of the principle of the invention; Fig. 2 is an explanatory diagram of the overall configuration of the embodiment; Fig. 3 is a schematic explanatory diagram of an image data processing module; Fig. 4 is an explanatory diagram of image data and vertical synchronization signals; Figure 5 is a diagram explaining the configuration of the feature quantity extraction module, Figure 6 is a time chart explaining the operation of the embodiment, Figure 7 is a diagram explaining the schematic configuration of the conventional system, 10 ● 12 ・ 14 ◆ 16 ・1 8 ・ 2 0 ● 22 ・ Module for processing image data Module for extracting feature quantities Counting means Counting means Setting means Storage means Changing means -13- -14- Horizontal synchronization signal Explanatory diagram of image data and vertical synchronization signal Fourth figure

Claims (1)

[Claims] A module (10) that processes each image data of continuous input.
and a module (12) for extracting feature quantities from processed image data are coupled in a pipeline, and the image processing module (10) includes means (14) for counting the number of input image data from a vertical synchronization signal. ), the feature extracting module (12) includes means (16) for counting the number of input images from the vertical synchronization signal, and internally calculating the number of images counted by the image processing module (10). A means (18) for setting the number of images corresponding to the waiting time based on the counted number of images, a means (20) for storing a parameter change command, and a means for setting the number of processed images input from the time of setting the number of images. An image processing system comprising: means (22) for changing parameters for feature extraction according to the stored command when the number of images reaches the number of images.