JPS6269384A - Automatic setting method for binary coding level - Google Patents

Abstract

PURPOSE:To cope with the variation of brightness with high flexibility by setting plural windows to calculate the mean luminance in each window and then calculating the optimum binary coding level according to said mean luminance. CONSTITUTION:In case the positions of parts are roughly decided, the windows 11 and 12 are set inside and outside the parts 21. Here the mean levels of luminance a window 12 inside parts and a window 11 outside parts are set at D1 and D2 respectively. Then these levels D1 and D2 are first calculated for both windows (inside and outside parts) and then the binary coding level S=D1+(D2-D1)XK is calculated (K: prescribed constant). In other words, the background is distinguished from the parts on the basis of the mean luminance of the window inside the parts and by means of the mean luminance difference between the background and the parts. Thus a binary coding level is calculated in response to the brightness despite the variation of the ambient brightness despite the variation of the ambient brightness since the mean luminance of the window inside the parts is used as a standard.

Description

Detailed Description of the Invention (Industrial Application Field) The present invention provides automatic setting of a binarization level, which is a threshold value when binarizing (white/black) video data from a TV camera, in a visual sensor. Regarding the method.

(Prior Art) Many methods for measuring the optimal binarization level are known. Among these, the fixed binarization method has been most commonly adopted. Since the binarization level is fixed, it cannot respond to changes in brightness at all. The other method is the floating binarization method, which binarizes depending on the changes in the video signal from the camera, so it can cope with changes in brightness to some extent, but the actual binary image obtained is not faithful. It lacks. For example, as a method for measuring the brightness distribution of each pixel of video data from a TV camera,
No. 9-114687, No. 59-140588, No. 59
-188786 and No. 58-184684 are known, and these methods, like the known methods such as the p-tile method and the mode method, calculate the optimal image based on the brightness distribution situation (histogram) of the image by various judgments. Although it obtains a value level, it is not suitable for processing complex images or different images.

Furthermore, as a result of measuring the characteristic quantities of video data from a TV camera, such as the area or center of gravity, several times while changing the binarization level, the binarization level that showed the optimal feature quantity was selected as the optimum one. The methods disclosed in Japanese Patent Application Laid-open Nos. 57-81687 and 59-229682 are known.
The problem was that the processing time was long. Furthermore, the method disclosed in Japanese Patent Application Laid-Open No. 148989/1989 is known as a method for visually selecting the optimum level by displaying it on a monitor, but this method measures all the brightness of the image to be processed and the background, and based on the results. However, since the measurement target is only one point on each image, errors may occur if there are variations between pixels or pixel defects, and judgments are made by humans. There are problems such as the inevitable errors associated with this.

(Problems to be Solved by the Invention) The present invention eliminates all the drawbacks of such conventional products.
Processing was performed using a binarization level that was set in advance as a threshold for binarization, but this value is normally set to -false, and the rest is fixed, so the brightness of the surroundings changes. If this occurs, it will adversely affect the video after binarization processing.

In the present invention, in order to cope with such a phenomenon and to reduce the change in the image after binarization processing even if the surrounding brightness changes, the visual sensor automatically adjusts the binarization level. The purpose is to realize a function to be corrected without having to go through visual selection by a human, and without requiring much force or long processing time.

(Means for solving the gap problem) Therefore, in the present invention, in a visual sensor that fully binarizes the image of the TV right camera and performs all image processing, a plurality of windows are all set, and the average brightness within the window is Calculate each of them and calculate the optimal binarization level from the results 'f! :! This is an automatic binarization level setting method that uses ¥f characteristics.

(Example) Embodiments of the present invention will be described with reference to all the drawings.

Generally, when working with all visual sensors, the positions of the parts to be visualized may be almost fixed, or they may be completely different. In the present invention, in order to correspond to each of these conditions, two types of automatic binarization level setting methods are implemented as described in 1, and can be used depending on the conditions.

A) When the position of the part is almost fixed and appears on the TV camera screen α0.

If the position of the part is roughly in the shape of a chitana/shime, first place it inside and outside of the part Q as shown in Figure 2.
Rewindow (11, 12) Set f. (Here, assume that the average brightness within the window α inside the part is fD, and the average brightness inside the outside window α is . outside) D+xDz? calculate. Next, the binarization level S is calculated using , D2 and Equation 1.

S=D, +(D2-D,)XK・・・・・・・・・l(
Here, K is a constant set in advance. ) Based on the average brightness of the window in the part (
(1st term), and further uses the average brightness difference between the background and the parts (2nd term) to distinguish the background from all parts. As a result, even if the surrounding brightness changes, the binarization level can be calculated according to the brightness because the average brightness of the window within the part is used as the reference.

B) The position of the parts is determined 1゛V force camera screen (if it does not appear in 10 parts.

In this case, a plurality of windows (11', 12', . . . ) are all set in advance at arbitrary positions as shown in FIG. In addition, the binarization level SR is set. First, take in the video of the reference part (21") and calculate the average brightness within each window. Then, store the maximum (or minimum) of these as the average brightness iDu of the window.Next When actually processing, import the image of the part (the position is different from when it was stored) and calculate the average brightness in each window '1ztt! and its maximum value (or minimum value) Dp
Find f. Then, by formula 2, the appropriate binarization level Bp
2 Calculate.

SF=sR+ (DP-DR)・・・・・・・・・
2 This means that if the average brightness of the background (or part) changes due to changes in brightness, consider the average brightness of the entire background (or part) in any window A, and carefully consider that it is the maximum (or part) brightness. By considering that the value of Dp -DR in the second term of Equation 2 becomes the degree of change in brightness, the binarization level SuK at the time of storage can be added (or subtracted) to the appropriate value. Binarization level Bpf
You can ask for it.

The automatic binarization level setting method is carried out by setting all the binarization levels obtained by the methods A) and B) in the binarization circuit (3) of the visual sensor shown in FIG. be able to.

(Effects of the invention) Generally, visual sensors use the TV left camera, which is sensitive to changes in ambient brightness, so special measures such as installing special lighting and blocking external light and other light are required. there were. This friend required various restrictions, special technology, and equipment, which created obstacles to using visual sensors.

However, with the present invention, it is now possible to flexibly respond to changes in brightness, and various obstacles can be overcome! ll will be reduced. Additionally, measurement errors and work errors caused by changes in the binarized image caused by changes in brightness are eliminated. Furthermore, it is now possible to automatically set the binarization level without requiring a long processing time.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of a visual sensor for carrying out the method of the present invention, and FIG. 2 and FIG. 3 show images of a TV left camera showing different embodiments of the method of the present invention. 1...TV camera 10, 101...TV camera screen 11, H', W, 11", 11", 11°...Window outside the part (second window) 12.12', 12 '...Window in part (first window) 20...Visual sensor 21.21°...Part

Claims (1)

[Claims] In a visual sensor that performs image processing by binarizing video data from a TV camera, a first window that displays the target part and a second window that displays the background of the target part are set on the TV camera screen, and each An automatic binarization level setting characterized in that the average luminance within the first and second windows is calculated, and a function of the difference between the results is corrected to automatically set the optimum binarization level. Method.