JPS6319080A - Printing quality estimating method - Google Patents

Abstract

PURPOSE:To estimate objectively printing quality in matching with the real sense of visual observation by displaying the extent of the unevenness of the outline of a printing image with the maximum amplitude value of the spread of a picture element. CONSTITUTION:The image of an original 1 is read through a microscope 2 by a television camera 3 and converted to an electric signal. The electric signal is stored through an A/D converter 4 to an image memory 5. A computer 6 obtains the regression straight line to the outline of respective areas of an image stored into the image memory 5 with a statistical method, and obtains, as a maximum amplitude value, the adding value of the distance from the regression straight line of an picture element to exist at the remotest position from respective regression straight lines out of the picture element to exist at the outside of the regression straight line obtained for respective sides and the picture element to exist at the inside. The maximum amplitude value is displayed at a CRT 8 as the estimating criterion of the printing quality and printed out by a printer 9.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a method for evaluating the quality of images printed by a printer or the like.

(Prior Art) In recent years, printers, especially printers as terminal devices, have become widespread and are used in all fields of industry. There are various types of printer wires, from high-end laser printers to simple thermal transfer printers.

In recent years, as printers have become more popular, interest in their image quality has also increased.

(Problems to be Solved by the Invention) Conventional methods for evaluating the printing quality of printers include, for example, in the case of thermal transfer printers, edge characteristics,
Criteria for judgment include the distortion of the printed pattern and the unevenness of the transfer, but the reality is that the actual evaluation relies entirely on visual sensation. Although some research has been conducted on the evaluation of print quality or image quality of printers, it is still at the stage of theoretical research and it is difficult to predict whether it will be of practical use.

The present invention has been made in view of these points, and
The purpose is to put into practical use a print quality evaluation method that can objectively evaluate print quality in accordance with visual perception.

(Means for Solving the Problems) The present invention, which solves the above-mentioned problems, uses a statistical method to find a regression line for the contour of each region of a sample image, and then calculates the regression line that lies outside the regression line found for each side. The summation value of the distance from the regression line of the pixel that exists at the point farthest from the regression line among the pixels that overlap and the pixels that exist inside and cover is determined as the maximum amplitude value, and this maximum amplitude value is used as the evaluation standard for print quality. It is characterized by the fact that it is used.

(Operation) The present invention uses a method based on the maximum amplitude value of pixel spread to evaluate print quality.

(Example) Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a flowchart showing an embodiment of the Gino method of the present invention. The following will be explained along this flowchart.

Step ■ Calculate regression lines for the contours of the top, bottom, left, and right sides of the sample image using a statistical method.

As a measure for evaluating the print quality of a sample image, the following can be considered: 1) a measure of edge unevenness (edge smoothness), 2) a measure of edge blur (edge sharpness), and 2) other measures. The method used in the present invention falls within the range of scales related to (1) and (1) edge unevenness.

Now, consider a printed image as shown in FIG. Normally, for example, a quadrilateral image as shown by the broken line in the figure should be printed, but due to printer performance constraints, the image is printed with ink only on the area shown by the diagonal line in the figure. It is assumed that it has been created.

The blank portions formed in the shaded areas indicate portions (referred to as voids) to which ink did not adhere.

Here, we will discuss a method for quantitatively evaluating the unevenness of the edges on the top, bottom, left, and right sides of the printed image shown in the figure.

FIG. 3 is an enlarged view of one side. As you can see in the figure, the outline of the sides is that of a rectangular pixel with each side being, for example, about 10 μm)! ! It is formed more than anything. The "10" shown in the pixel indicates the center point. Now, for each pixel, find a straight line (referred to as a regression line) that passes through a distance close to the center point of each pixel.

Statistical methods are used to find the regression line, and here the least squares method, which is the most common method, is used.

To find a regression line using the method of least squares, when a set of n pixels (xi, yi) (i = 1 to n) is given, a linear equation such that the data of these pixels also fits green well is calculated. All you have to do is ask for. Therefore, the linear equation to be applied is y=ax+b...(1
), and set the coefficient a so that it most closely matches the actual measured value.

Determine b.

Now, as shown in FIG. 4, an arbitrary pixel (coordinate x i.

yi) is given, the coefficients a and b are determined so as to minimize the sum of squares of the difference in distance between the measured value yi and the theoretical value y, that is, the sum of squared residuals. Since the residual of point (xi, yi) is yi − (axt +b >, the residual sum of squares 1
(a, b) is given by the following equation.

1(a,b)-Σ(yi-(a+bxi)')
2I-1... (2) However, n is the coefficient a, l) that minimizes the number of sample pixels 1 (a, t+), by polarizing I (a, b) by a and b. Differentiate and set it as zero. From al(a, b)/aa=0, <a/aa) Σ (■i-a -bxi
)? =Q... (3) 1+1 a E < a , b )/ah = 0, so (a/ab)Σ(yi-a -bxi) 2=O-<
4) From equations (3) and (4), 2Σ (yi-a -bxi) x (-1
>=O-(5)1+1 2,," (yi-a-bxi) x (-xi)
=O...-<6)1+1 If we rearrange using the transfer term L, we can obtain the following simultaneous equations.

(Σxi) a + (Σxi2) b = -Σxiyi
-(8): ji-11-+ Solving <7) and <8) for a and b yields a and b, and the regression line equation shown in equation (1) is determined.

Figure 5 shows the regression line obtained in this way.
L in the figure is the regression line. l between each pixel that constitutes the contour and the regression line 1-! , fil+, . . . are the closest ones when considered on average. That is, the one that best approximates the position of each pixel is selected.

Step■ The maximum amplitude value is the sum of the summed values from the regression line of 1ilii elements that are located at points that are green away from the regression line among the pixels that exist outside and the pixels that exist inside the regression line found for each side. demand.

This will be explained using FIG. When the regression line 1- is determined, the pixels forming the contour line are divided into two, pixels existing outside and pixels existing inside, using the regression line as a boundary line. Among the pixels existing on the outside and covered, focus is placed on the one that is farthest from the regression line 1-, and the distance from the regression line [- is set to 4011. Next, focus on the regression line that is farthest from the regression line of the pixels existing inside, and set the distance from the regression line 1 to 12 m2. After finding 7111 and 1!1l12, add value of /l11 and 12m2<l! m++(?
I02) and calculate this as the maximum amplitude value (peak to
peak value). The maximum amplitude value Vp is given by the following equation.

\/ li = , (l m l+ / nl 2
...<9) Step ■ The obtained maximum amplitude value is used as the evaluation standard for printing quality.

As mentioned above, the maximum amplitude value can be used as a measure of the degree of unevenness at the edges of an image. That is,
When the maximum amplitude 1 straight ■p is large, the degree of unevenness is large, so it can be determined that the print quality at the edge of the image is poor, and when the maximum amplitude value Vp is small, the degree of unevenness is small, so the print quality at the edge of the image is good. It can be determined that

Next, experimental results of print quality evaluation using the present invention will be explained. FIG. 6 shows a sample image used in the experiment, which is displayed within a frame of 450 pixels each in the vertical and horizontal directions. The sample image shown in the figure is obtained by reading the actual printed image with an image reading device.
It shows a binary image that has been converted into a value. This simple threshold for binarization (
PoiTh used the following formula.

A - dh = 0.6 <t ≧ 0.6) t (t <0.6> ... (10) However, (= (dw + db> / 2dW; paper reflection m
db;Reflection density of the image area shown in Figure 6.2tlf+ For each side of the top, bottom, left and right of the image, cut 10% of both ends, and apply the present invention to the remaining 80%)n. When the maximum amplitude value was calculated, the upper side was
The bottom side, left side, and right side are each 0.409 (mm>,
0.612 (mm>, 0.280 (mm)
.

It became 0.276 (mn+). As is clear from FIG. 6, the print quality is worse on the upper and lower sides than on the left and right sides, but the experimental results also agree well with this.

FIG. 7 is a diagram showing an example of the configuration of a hardware system for implementing the method of the present invention. The image recorded on the original 1 is optically magnified by a microscope 2 and then viewed by a television camera 3.
is read and converted into an electrical signal. The image signal converted into an electrical signal is converted into digital data by an A/D converter 4. Here, the number of bits of the A/D converter 4 may be as long as it has a resolution of about 256 gradations to discriminate between gray levels. That is, an A/D converter of about 8 bits may be used.

The output of the A/D converter 4 is sequentially stored in the image memory 5 pixel by pixel. Here, the memory capacity of the image memory 5 has a capacity large enough to store all the images recorded on the original 1, for example, 512 x 640.
Vixel level is appropriate. The computer 6 sequentially reads out the image data stored in the image memory 5, performs binarization processing, and stores 2111i data for each pixel on the floppy disk 7. Here, the image of the binary image stored on the floppy disk 7 is as shown in FIG.

Next, the computer 6 reads the back, top, bottom, and left sides of the data stored on the floppy disk 7.

Only the data forming the contour of the right side is read out, and a regression line is determined by the least squares method. Once the regression line is determined, the maximum amplitude value of pixel spread is determined using this regression line as a reference. The maximum amplitude value determined for each side is displayed on the CRT' 8 or printed out on the printer 9. The keyboard 10 gives various commands to the viewer 6 .

In the above description, the least squares method was used to find the regression line, but other statistical methods may be used. The printed image used in the present invention may be printed by any type of printer such as thermal transfer, inkjet, wire dot, etc., and the present invention is equally applicable to these printed images.

In the above-described embodiment, the case where the maximum amplitude value of pixel spread is determined for the binary image shown in FIG. 6 was explained as an example. However, the present invention is not limited to this. For example, for the binary image shown in Figure 6, after printing inside the binary image, the area (void) is filled in, and then the pixels of the overhanging part of the outline are printed in the direction of the center of the image. Move parallel to the x-axis (or y-axis) to lower the overhang. The present invention can also be applied to binary images in which voids and overhangs are eliminated in this manner.

(Effects of the invention in Song Dynasty) As explained in detail above, according to the present invention, the degree of unevenness of the outline of a printed image is expressed by the maximum amplitude value of pixel spread1, thereby providing an objective that matches the actual feeling of visual inspection. It is possible to realize a printing quality evaluation method that can evaluate printing quality.

[Brief explanation of the drawing]

FIG. 1 shows a flowchart showing an embodiment of the method of the present invention.
Figure 2 is a diagram showing an example of a printed image, Figure 3 is an enlarged view of the outline, Figure 4 is an illustration of the least squares method, Figure 5 is a diagram showing a regression line, and Figure 6 is a diagram used in the experiment. Diagram showing an example of a binary image, 7th
The figure is a diagram showing an example of the configuration of a hardware system that implements the present invention. 1... Manuscript 2... Microscope 3...
TV camera 4...A/D converter 5...Image memory 6...Computer 7...Floppy disk 8...CRT9...Printer
10...Keyboard care 1 Figure 2

Claims (2)

[Claims] (1) A regression line for the outline of each region of the sample image is calculated using a statistical method, and the point farthest from the regression line among the pixels existing outside and inside the regression line calculated for each side. A print quality evaluation method characterized in that the sum of the distances of pixels existing in the regression line from the regression line is determined as a maximum amplitude value, and this maximum amplitude value is used as a print quality evaluation standard. (2) The print quality evaluation method according to claim 1, wherein a least squares method is used as the statistical method.