JPS60200691A - Binarization system - Google Patents

Abstract

PURPOSE:To enable binarization stably by correcting a threshold stored in memory in accordance with the difference between data on a calculated average level and stored data on an average level and setting it as a next threshold in terms of image pickup device. CONSTITUTION:When an operator decides an appropriate threshold (T) viewing a screen of display device 12, the operator stores it in a memory 14, and simultaneously calculates an average W of a photographing signal level corresponding to a window area 22, which is set in the backward in a visual field 20 of an image pickup device 1 in terms of software. Then a reference deviation epsilon, that is, the difference between the average W and the finally decided threshold T is calculated by a microcomputer 7. The threshold T is substracted from the obtained average W, whereby the reference deviation epsilon is calculated and stored in the memory 14. The threshold is corrected with aid of the epsilon, which is made into the next threshold. Thus a threshold can be automatically optimized by following the influence which is affected to an input picture of an image pickup device by a change in the outside brightness.

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field of the Invention The present invention relates to a binarization method for binarizing an imaging signal obtained by imaging with an imaging device, and particularly relates to an improvement in the method of setting the dark value.

Conventional technology and problems In visual sensors that process binary images to detect the characteristics of objects, the darkness value when binarizing the image signal is an important factor for correctly analyzing the image. In places where a stable and strong light source cannot be obtained, stable binarization is difficult if the darkness value is fixed because it is affected by changes in external illuminance.

For this reason, various methods have been proposed to make the dark value a floating level. For example, it is known that the average level of the entire screen is calculated and the darkness value is corrected according to fluctuations in this average level, but the screen contains a detected object that you want to binarize. Even if the same darkness value is sufficient, the average level of the screen may change due to changes in the proportion of the object to be detected in the visual field, which has the disadvantage that it is not possible to accurately correct the darkness value. .

Purpose of the Invention The present invention aims to improve these conventional drawbacks.
The purpose of this is to enable stable binarization even if the illuminance of the outside world changes and even if the size of the object to be detected changes.

Structure of the Invention The gist of the present invention is to provide a binarized image obtained by capturing an object to be detected within the field of view of the imaging device in a binarization method that binarizes an image signal obtained by imaging with an imaging device. While observing on a display device, the threshold value T that was appropriately set and the data regarding the average level of the imaging signal corresponding to the window area set in the background part within the field of view at that time are stored in memory C, and each of the following 2 For each value conversion process, data regarding the average level of the imaging signal corresponding to the window area at that time is calculated, and data is stored in the memory according to the difference between the calculated data and the data regarding the average level stored in the memory. This is a binarization method in which the stored threshold value T is corrected and set as the next darkness value.

When capturing an image of an object with an imaging device, it is common for the object to be positioned approximately in the center of the imaging field of view.
A background portion that does not overlap with the object to be detected generally exists in the corner of the field of view of the imaging device. The brightness of this background portion changes in the same way as the brightness of the object to be detected, depending on the brightness of the outside world. Therefore, data regarding the average brightness level of the background area when the optimal threshold value T is set is stored, and data regarding the average brightness level of the background area is detected again during subsequent binarization processing. By comparing this with the memorized average level, you can see how much the brightness has changed, so if you change the threshold T in accordance with that change, the change in external brightness will affect the input image of the imaging device. It becomes possible to automatically optimize the darkness value by following the influence. and,
Since the darkness value is corrected based on the average level of the background area where the object to be detected does not exist, it is naturally not affected by the size of the object to be detected.

Embodiment of the Invention FIG. 1 is a block diagram of a main part showing an example of a device implementing the present invention, and shows the configuration of a visual sensor processing device that notifies an industrial robot 1 of information such as the position of an object to be detected. . In the figure, l is an analog imaging signal from an imaging device such as an ITV camera, which is input to a converter 2, linearly converted into a digital imaging signal at a predetermined sampling period, and then binarized. The circuit 32 is input to the gradation image memory 4. The output number of the single Al converter 2 is, for example, 6 bits, and therefore the level of one pixel is expressed as a digital value of 64 gradations. Each time the binarization circuit 3 compares the level of the imaging signal and the dark value C set in the latch circuit 5, for example, when b is larger than C, the output is set to "1",
When it is small, the output is set to "0". Binarization circuit 3-2
The digitized signal is input to the binary image memory 6. When an image capture command is input from the microcomputer 7 to the address generator 9 via the buffer 8, the address generator 9
2 at the timing based on the synchronization signal from the imaging device l.
Generates addresses for the value image memory 6 and gradation image memory 4, and imports one screen worth of binary image and 1 gradation image into the binary image memory 6 and gradation image memory 4, respectively. The computer 7 processes the binary image captured in the binary image memory 6 to detect the position and orientation of the object to be detected, and notifies this to an industrial robot (not shown) via the interface circuit 10. 11, the contents of the binary image memory 6 can be displayed on a display device 12 such as a CRT.
can be displayed.

Now, the binarization method according to the present invention is implemented as follows.

First, as shown in FIG. 2, for example, the field of view 2 of the imaging device l
The object to be detected 21 is placed within 0. After setting an appropriate darkness value in the latch circuit 5 using the keyboard 14, the microcomputer 7 is commanded to capture an image, and the contents of the binary image memory 6 are displayed on the display device I2.

The operator looks at the screen of the display device 12 and reads 'd! , observe whether the detected object 21 portion is properly binarized;
If the image is not binarized properly, changing the settings of the latch circuit 5 and capturing and displaying the image 7 are repeated until the image is binarized properly.

Once the appropriate darkness value (referred to as T) is determined in this way,
This threshold T is stored in the memo 4, and the average value W of the imaging signal level corresponding to the window area 22 set by software in the background within the field of view 20 of the imaging device 1 as shown in FIG. is calculated, and the standard deviation ε, which is the difference between this average value W and the finally determined threshold value T, is made to be calculated by the microcomputer 7 as data regarding the average level of the background portion. To calculate the average value W, the microcomputer 7 sequentially reads out the contents of the addresses of the gradation image memory 4 in which the imaging signals corresponding to the window area 22 are stored.
This is done by taking the average value of this, and the average value W
Calculate the standard deviation ε by subtracting the threshold T from
Memorize this on January 4th.

The above processing is performed before actually operating the visual sensor processing device, and may be performed, for example, during a teaching mode in which the visual sensor processing device is taught the characteristic values of the detected object.

When the visual sensor processing device is put into the execution mode, when an image processing command is received from the industrial robot via the ink face circuit 10, a binary image for one screen of the imaging device 1 is captured in the binary image memory 6, and A gradation image for one screen is taken into the gradation image memory 4, and by processing the binary image, the position of the object to be detected, etc. is detected, and processing is performed to notify the industrial robot. As the binarized dark value at this time, a value previously set in the launch circuit 5 is used. Also, the micro convenience store 7 is the window area 2 at this time.
The contents of the addresses of the gradation image memory 4 in which the imaging signals corresponding to 2 are stored are sequentially read out, the average value W' is determined, and the reference displacement ε stored in the memory 14 is added to this using the following formula. Calculate the new darkness value T゛ and latch it once
Set to 15.

T' = W' + ε -11) When the image processing command is input again from the ink face circuit 10, this time it is binarized with To as the dark value, and after this binarization process is finished, the window area at this time is again The contents of the addresses of the gradation image memory 4 in which the imaging signals corresponding to 22 are stored are sequentially read out, and the average value W'' is calculated.
From this and the reference displacement ε recorded in the memory 14 as 1 and a, a new threshold value 'r'' is calculated by the following equation and sent to the latch circuit 5.

′r”−W”+t −(2) Instead of storing the standard deviation ε, the average value W of the window area is stored as data regarding the average level, and the next threshold value ′ at each binarization process is stored. The calculation of r° may be performed using the following formula.

T” −T+ (W′ −W) −(3) However, W′ is the average value.As explained in detail in this invention, according to the present invention, changes in the brightness of the outside world affect the input image of the imaging device. It is possible to automatically optimize the darkness value by following the influence, and the darkness value is corrected based on data regarding the average level of the background area where there is no object to be detected, so it is not affected by the object to be detected. Furthermore, in order to achieve higher accuracy, first import a gradation image, calculate the average level of the background area, correct the darkness value,
You can import a binary image using this corrected darkness value, but it takes time to obtain the results because two image import processes are required for each binarization process. There are too many drawbacks.

In contrast, according to the present invention, at each binarization process, the darkness value corrected during the previous binarization process is used to perform the binarization, and at the same time, the next darkness value is set. Therefore, in each binarization process, only one image capture process is required, and the time from receiving a binarization process command to obtaining a result can be sped up. Generally, in vision sensors for industrial robots, binarization processing is performed every short period of time, and the brightness of the outside world does not change so rapidly in a short period of time, so the darkness value detected last time is used as in the present invention. However, binarization can be performed reliably. Of course, the effects of the present invention are best achieved when the brightness of the outside world changes slowly over a long period, such as when driving all night.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of essential parts showing an example of an apparatus implementing the present invention, and FIG. 2 is an explanatory diagram of its operation. I is an imaging device, 2 is an A/D converter, 3 is a binarization circuit, 4
5 is a gradation image memory, 5 is a launch circuit to which the dark value is input, 6 is a binary image memory, and 7 is a microcomputer. Patent applicant FANUC Co., Ltd. Representative Patent Attorney Tamamukyu 2 people outside the department Figure 2

Claims (1)

[Claims] In a binarization method that binarizes an image signal obtained by capturing an image with an imaging device, a threshold value that is appropriately set while observing a binarized image of an object to be detected within the field of view of the imaging device on a display device. T and data regarding the average level of the imaging signal corresponding to the window area set in the back area within the visual field at that time are stored in a memory, and the window area at that time is stored for each subsequent binarization process. data regarding the average level of the imaging signal corresponding to is calculated, and the threshold T stored in the memory is corrected according to the difference between the calculated data and the data regarding the average level stored in the memory, A binarization method characterized in that it is set as a value.