JPS619768A - Peripheral distribution calculator - Google Patents

Abstract

PURPOSE:To reduce the effect of a background and to obtain the clear features of pictures by comparing the picture element data with the background level and totalizing successively only the horizontal or vertical direction. CONSTITUTION:The picture image signal obtained by scanning the photoelectric conversion surface of an ITV camera 11 is stored in a rescribed address of a picture memory 13 after A/D conversion. A comparator 15 compares the picture element data with the background level held by a latch 20. Then an adder circuit 16 is closed only in case the input picture element data is larger than the background level. Then the picture data sent from an A/D converter 12 is added with the data held by a latch 21 and then held again by the latch 21. This operation is repeated to obtain the peripheral distributions SIGMAXnot equal to , SIGMAX2- of the direction X. The peripheral distributions of the direction Y can also be obtained in the saem way. Thus it is possible to reduce the effect of the background level and to obtain the clear features of pictures.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to an image recognition device, and more particularly to a marginal distribution calculation device that detects image features from the marginal distribution of a two-dimensional image signal.

[Technical background of the invention and its problems]

An image signal sent from a two-dimensional photoelectric conversion device such as an ITV camera is A/D converted into pixel data of a predetermined gradation, and the A/D converted pixel data is accumulated in the horizontal and vertical directions. There are ways to recognize images.

That is, if the image obtained from, for example, an ITV camera is 4 pixels x 4 pixels, the image signal is 25 of O~255.
Assuming that the pixel data shown in FIG. 3 is stored in the image memory (1) after being A/D-converted into 6 gradation levels, the X-direction peripheral distribution Σ is obtained by adding the pixel data in the X-direction. X is 37.45,83°33 from the top. Also, the Y-direction peripheral distribution Σ added in the Y-direction. Y is 33.50,7 from the left
It becomes 7.38. Therefore, in this image recognition method, these X-direction and Y-direction peripheral distributions Σ. From X, Σoy,
As shown surrounded by a solid line (2) in the drawing, it is possible to detect a bright feature approximately in the center of the image. Also, these marginal distributions Σ. X, Σ. By calculating the average value of Y, the center of gravity or a value close to the center of gravity can be found.

However, in general, images obtained from photoelectric conversion devices such as TV cameras include noise and level shifts in the image signal caused by the background and characteristics of the photoelectric conversion device.
In the example shown in FIG. 3, the background level is 5 to 10. Although the image features are still somewhat obscured in the example shown in Figure 3 due to this background level,
If this is distributed over the entire surface of a larger image, such as 256 pixels x 256 pixels, or if there is not much difference between the feature level and the background level, the accumulation of background levels cannot be ignored. Tonashi,
The peripheral distribution, which should originally be a characteristic, is obscured and cannot be recognized.

[Purpose of the invention]

An object of the present invention is to obtain a marginal distribution calculation device that can reduce the influence of the background and clarify the features of an image.

[Summary of the invention]

The image signal sent from the two-dimensional photoelectric conversion device is A/D converted into pixel data of a predetermined gradation, addressed and stored in memory, and this addressed pixel data is obtained from the two-dimensional photoelectric conversion device. Compared with a background level determined from the background of the image being displayed, only pixel data larger or smaller than this background level is sequentially accumulated in the horizontal or vertical direction, so that the features of the image can be made clear. I made it.

[Embodiments of the invention]

Hereinafter, the present invention will be described based on embodiments with reference to the drawings.

FIG. 1 shows an embodiment of the marginal distribution calculation device of the present invention. This peripheral distribution calculation device converts an image signal sent from a two-dimensional photoelectric conversion device such as an ITV camera αD installed so as to face a subject (10) into a predetermined gradation, for example, 2.
A/D conversion to pixel data with 56 gray levels 8
The bit A/D converter (13) is connected to the comparator (L9) and the adder αυ via the image memo!J (131) and the switch α.

The image memory α East is also connected to an address generator α operated by a counter provided in the control device (L8).
→, and the pixel data A/D-converted by the A/D converter a can be stored at a predetermined address in the image memory 03 by address signals sent from the address generators α. This image memory (1
The output side of 3 is also connected to the comparator cl■ and the adder αe, and by the addressing of the address generator (19), the pixel data accumulated in the 90th floor of the image memo is sent to the comparator (I9) and the adder αe. The comparator (151) is also connected to a latch that holds the background level sent from the control unit (1B), and the output from this comparator (alpha) allows the A/D Inputs from the variable transformer α2 and the image memory αυ to the adder α0 are opened and closed.A latch (2I) for holding accumulated pixel data is connected to this adder (1e). Adder αe adds new pixel data input from A/D converter α4 or image memory (131) to the data held in latch Qυ.The added data is held in latch Qυ again. When the adder <161 completes the predetermined accumulation, the controller α reads the accumulated data based on the command obtained from the counter.

Assume now that the ITV camera αη has a resolution of 256 pixels x 256 pixels, scans its photoelectric conversion surface in the horizontal direction (X direction), and scans in the vertical direction (Y direction) without interlaced scanning. Assuming that the memory (I3) has a storage capacity of 256 pixels x 256 pixels, first
The image signal obtained by scanning the photoelectric conversion surface of the T■ camera aυ is A/D converted into pixel data of a predetermined gradation by an A/D converter (1), and then converted to pixel data of a predetermined gradation. Turn (1
An address generator (in hexadecimal notation) that generates an address signal (in hexadecimal notation) is stored at a predetermined address in the image memory (by way of addressing).At the same time,
The address generator (11) closes the switch (14) according to the address signal, thereby causing the image memory (
The same pixel data as the pixel data input into 13 is sequentially sent to comparator 09 and adder αe. Comparator Q
compares this pixel data with the background level (described later) held in the latch, and only when the input pixel data is greater than the background level (the pixel data is brighter than the background level), the adder (circuit 161) By this closing, the adder (16) performs addition of the pixel data sent from the A/D converter and the data held in the launch Cυ.Then, the added data is held by the latch (21) again.

In this way, when the counter provided at the control device α advances from 0 to 255, the latch Qυ holds the peripheral distribution ΣIXI in the X direction in which Y, which is the sum of only pixel data larger than the background level, is 1. Ru.

Then, this X-direction peripheral distribution Σ1X1 is read by the control device based on the instruction obtained from the counter, and at the same time, the latch (21) is cleared to 0.

The marginal distribution in the X direction with is the same from now on, and Y is 2,
3. Σ1X2. Σ, X, .

Next, the marginal distribution ΣlY in the Y direction is determined. This peripheral distribution ΣlY in the Y direction is first generated by the second signal generated from the address generator 4 following the address signal shown in FIG. 2A.
This is done by opening the switch (141) in response to the address signal shown in Figure B, and sequentially sending out the pixel data stored at a predetermined address in the Q floor of the image memory to the comparator (lω).The calculation method is as follows: Except for supplying pixel data from the image memory (13), everything is the same as the calculation of the peripheral distribution in the X direction, so detailed explanation will be omitted.

The background level used in calculating the peripheral distribution in the X direction and Y direction is determined in advance by appropriately extracting the level corresponding to the background part of the image from the pixel data obtained from the A/D converter. It will be done. Alternatively, if the pixel data obtained from the A/D converter α2 has a normal distribution or a distribution close to it, the value of 3δ obtained by statistically processing the pixel data is used as the background level. etc. can be arbitrarily determined.

As mentioned above, the A/D converter 82) or the image memory (1
By comparing the pixel data sent from the comparator (15) with the background level and adding only the pixel data that is higher than this background level, the influence of the background level is reduced and bright parts of the image are In other words, the features of the image can be made clear.

Next, other embodiments will be described.

In the embodiment described above, pixel data larger than the background level is accumulated when calculating the peripheral distribution, but depending on the case, pixel data smaller than the background level may be accumulated. Further, when comparing the pixel data with the background level, the pixel data may be subtracted and then accumulated.

Furthermore, in the above embodiment, the data value of the marginal distribution after the accumulation is read by the control device each time, but this is done by connecting the memory to the adder or the latch that holds the data value after the addition. It may be stored in this memory.

Furthermore, in the above embodiment, the pixel data was addressed by an address generator, but if time constraints are ignored, the control device can access the pixel data directly, and the sequence for determining the peripheral distribution shown in the above embodiment can be performed entirely by software. It's okay.

In addition, in the above embodiment, the peripheral distribution in the X direction is determined by sending pixel data directly from the A/D converter to the comparator and adder, but this peripheral distribution in the X direction is also the peripheral distribution in the Y direction. Similarly, the calculation may be performed using pixel data once stored in the image memory. In this case, the switch shown in the above embodiment becomes unnecessary.

Furthermore, in the above embodiment, an 8-bit A/D converter was used and the image memory was configured with a storage capacity of 256 pixels x 256 pixels, but the marginal distribution calculation device of the present invention has the following features:
It is not particularly limited to these.

〔Effect of the invention〕

The pixel data that has been A/D converted to a predetermined gradation is compared with the background level in a comparator, and only the pixel data that is higher than the background level or the pixel data that is lower than the background level is accumulated to calculate the marginal distribution. This makes it possible to reduce the influence of the background level and make the characteristics of the image clear.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the present invention, and FIG. 2 is a diagram showing the pattern of address signals sent from the address generator. FIG. 3 is an explanatory diagram of a conventional method for calculating the peripheral distribution. α0)...Subject (1])...ITV camera (1)...A/D converter (1)...Image memory α...Comparator (161...Adder α) ...Control device I...Address generator, c!η...Latch

Claims (1)

[Scope of Claims] A device that A/D converts an image signal sent from a two-dimensional photoelectric conversion device into pixel data of a predetermined gradation, a memory that stores the A/D converted pixel data; A circuit that sequentially addresses the pixel data in the memory in the horizontal or vertical direction; and a circuit that compares the pixel data addressed by this circuit with a background level determined from the background of the image obtained from the two-dimensional photoelectric conversion device; A peripheral distribution calculation device comprising: a circuit that sequentially accumulates only pixel data larger than a level (or pixel data smaller than a level) in a horizontal or vertical direction.