JPS6376578A - Automatic binarization system - Google Patents

Abstract

PURPOSE:To faithfully reproduce the targeted edge line of an input image in a binary image by calculating, from illuminance distribution, dots that are to form the edge line, and binarizing the illuminance level generated from these points to make it a threshold. CONSTITUTION:The illuminance distribution on one scanning line of an image in a window 16 is obtained, and such a range x1-x2 whose illuminance variation is maximum is calculated. Within the range x1-x2 the absolute values off the differences between neighboring picture elements are obtained; a maximum illumination variation point is obtained through the weight average processing of the results of the above-mentioned operation, and the illuminance level f' of thus obtained maximum variation point x' is referred to. This processing is repeated n-times from plural scanning lines to obtain the average value of the illuminance levels of thus n-times-obtained maximum variation points, and the result is made a threshold. As a result, the target edge line of an input image can be faithfully reproduced even in case when the target edge line includes a variation due to such as thin part or blurred part, or is influenced by outdoor daylight, etc.

Description

[Detailed Description of the Invention] [Industrial Field of Application] The present invention relates to binary image processing, and the effects on binary image conversion include lighting conditions, printing conditions, material of the object, etc. Automatic binarization method suitable for objects to be recognized under conditions that vary depending on the object, such as color [Prior art] The conventional method uses one scan as described in Japanese Patent Laid-Open No. 57-184368. Create a histogram on the line, the occurrence counternumber or peak island, P! The threshold value is set by finding the pixel information level where the frequency of appearance of that section is the minimum value. However, it is good if the pixel information level with the appearance frequency of the peaks P, , P can be detected, but the printing quality is poor, and both the background and the printed part are damaged! #If there are variations in the elementary information level, it is difficult to detect P,,P, where the coal output degree reaches its peak.

[Problems that the invention attempts to solve]

The above conventional technology can be applied to paper with good printing quality, but the surface of ICs, etc. that are the targets of screw stems has a matte finish, and the printed characters and background also have variations in brightness.
Furthermore, with low contrast, the characters become distorted, such as blurring, blurring, etc. The object of the present invention is to set an optimal threshold value for the above-mentioned object and faithfully reproduce the object edge of the input image into a binary image.

[Means for solving problems]

The above purpose is to calculate the maximum brightness change point at which the change in brightness on one scanning line is maximum, and use the brightness level of this maximum change point as a threshold to faithfully match the target edge line of the input image. Can be reproduced during binarization.

[Operation] This method directly calculates the point that should become the target edge line from the brightness distribution on one scanning line, and then binarizes the edge line using the brightness level of this point as a threshold. [Embodiment] An embodiment of the present invention will be described below with reference to FIG. 2. Second
The figure shows an example of a system implementing the invention. The system includes an imaging camera 1. 2. An illumination light source that uniformly illuminates the object 8 such as Ic. A magnifying lens 9 for magnifying the object 8. An XY stage 10 for correcting the position of the object 8. 3. A monitor that displays the captured original image or the binarized image. Based on data from camera 1 Il! ll1l! an image processing device 4 for processing, a CBr 4 for displaying operation details; It consists of a floppy disk (F'D) drive 6 for storing data and a keyboard 7 for inputting various commands and data.

In the above configuration, the image of the product name of the ICa appropriately positioned by the XY stage 10 is inputted to the image processing device 4 by the camera 1, and is stored in the FD in the FD drive 6 by operations using the C-T5 and the keyboard 7. By activating the evaluation program, the product name of Engineering C8 is recognized.

The results of the original image and the binary image are output to the monitor 3.

FIG. 3 shows the hardware configuration of the image processing device 4 of this system. The image processing device 4 includes a video input/output section 12 that inputs an image from a camera 21 and outputs the image to the monitor television 3, and an area division circuit section 11 that analyzes the input image and divides the input image into areas according to white and black clusters. A CPU section 1 (upper section 6) performs overall control such as operations from the C-T5 and continuous output of evaluation programs from the FD drive 6. As a result, the image input from the camera 1 is input to the video input/output unit 12, and
The image is binarized according to an instruction such as Br4, and the binary image is analyzed by the area dividing circuit s11 and further recognized by the evaluation program stored in the FD drive 6 by the CPU section.

Next, FIG. 4 shows a 181I image in which 1080 product names have been input. The white part is the IC product name printing part 15. The black part is the background 14. In reality, the input image in Figure 4 is 0~
It is expressed in 255 to 256 gradations.

When the input image of the IC 8 is binarized, even if it is the same object, there are changes as shown in FIG. 5 due to the influence of illumination and outward direction. The image in the window 16 in FIG. 4 will be explained as an example. The brightness distribution on one scanning line 17 of the image within the window 16 is shown in FIG. Even for the same object, due to the influence of extraversion, the value is 18 when the contrast is high and 19 when the contrast is low. In such a case, binarization using the same threshold value is difficult, and binarization using an appropriate threshold value is required.

In particular, as with the target of this system, the area used to determine the conventional threshold value is Identification using parameters such as the concavity length of one circumference is difficult, and the edge deflection of the object after binarization determined using the above parameters is not faithfully reproduced. Therefore, in this system, we have developed a binarization method that faithfully reproduces the target edge line using the method described below.

A schematic diagram thereof is shown in FIG. 6, and a flow thereof is shown in FIG. 1 and will be explained below.

First, input the image (Sl) Figure 6 (a), (b)
As shown in (1) the brightness distribution on one scanning line of the one-plane image within the window 16 is taken (S2).

(2) As shown in FIG. 6(b), calculate the section (x, to x snow) where the change in brightness is maximum. (S5).

(3) Line 5 (84) performs differential X-minute processing (takes the absolute value of the difference in brightness with an adjacent pixel) in the section (z, to x,) where the brightness change is maximum. The results are shown in FIG. 6(C).

(4) The X coordinate value resulting from the differential processing is weighted and represented by the following formula, and the point of maximum brightness change is determined by weighted averaging processing (S5).

Weighted average,
fL-1) Brightness maximum change point:: Brightness level at X coordinate: (S6).

A method of referencing the brightness level is shown in FIG.

The obtained maximum change point (=) is the pixels before and after it (xl).

xt), and assuming that the brightness and coordinates are linear in the minimum, the pixels before and after the maximum change point (=) (x
', , x', ) between the brightness levels f'+*f'x, r: 1 to -Z': ! ; The brightness level 7 of the point divided at the same ratio as −J is set as the brightness level of the maximum change point.

(6) Perform the processes of (1) to (5) twice using multiple scanning lines (S7), calculate the average value p (= (S8). This determines a binarization threshold that can faithfully reproduce the edge line of the target in the input image even if the target has changes such as blurring or blurring, and is also affected by externalization. be able to.

(Effects of the Invention) According to the present invention, a point that should become an edge line of interest is calculated from the brightness distribution, and the brightness level of this point is set as a threshold value of 2.
By converting into a value, the target edge line of the input image can be faithfully reproduced into a binary image.

[Brief explanation of the drawing]

FIG. 1 is a schematic flow diagram of an embodiment of the present invention, FIG. 2 is a configuration diagram of an embodiment of the invention, FIG. 3 is a hardware configuration diagram of an embodiment of the invention, and FIG. 4 is a diagram similar to the one shown in FIG. An explanatory diagram of the fcs input image, Figure 5 is a brightness distribution diagram in the window 16 in Figure 4, Figure 6 is a schematic diagram of the present invention, and Figure 7 is an enlargement of the maximum change section of Figure 6 (b). Figure 016: Window, 17: Scanning line, 18.19: Brightness distribution, 1: Camera, 2: Lighting, 6: Monitor television, 4: Image processing device, 5: CRT. 6...FL) drive, 7...keyboard. 8...IC, 9...Magnifying lens, 10...XY stage.

Claims (1)

[Claims] 1. In the threshold determination method in binary image processing, find the section where the change in brightness on one scanning line is maximum, perform differential processing within that section, and calculate the maximum brightness change point by weighted averaging of the results. The automatic binarization method is characterized in that the brightness level at the point of maximum change is referred to, these steps are repeated for a plurality of scanning lines, and the average value is used as the threshold.