JP2001243473A - Image density unevenness detection method - Google Patents

Abstract

(57) [Problem] To provide an image density unevenness detection method for improving detection sensitivity of dot-like unevenness in color filter unevenness detection. An average luminance value (L _AP ) of all pixels located in an area (AP) occupied by Wx and Wy pixels is prepared from pixels of a two-dimensional image, and 2) an outer side of a D pixel from the area (AP). An average luminance value (L _AS ) of all pixels located in the donut-shaped area (AS) having a width of H pixels is prepared, and 3) a difference between the average luminance value (L _AP ) and the average luminance value (L _AS ) is calculated. Detecting a non-uniform area by comparing with a non-uniformity detection threshold.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the detection of a non-uniform area in which a partial density generated on an inspection object which is expected to have a constant density throughout the entire inspection area is analyzed and detected from the photographed image. More particularly, the present invention relates to a method for detecting image density unevenness used in an unevenness inspection of a color filter or the like.

[0002]

2. Description of the Related Art As a part of quality inspection of a color filter for color display of a color liquid crystal display device and a color filter for color separation of a color video camera, unevenness inspection using an image analyzer is widely performed. This image analysis device usually consists of an image input device, a computer, lighting, and attached devices. After obtaining image data by photographing a color filter with transmitted light, the image processing is performed by a computer, and an uneven area is specified. are doing. As an image processing method used for this, for example, after performing a digital filter, a secondary differentiation process, etc. for each pixel of image data,
In comparison with a separately set threshold value, a portion exceeding the threshold value is specified as an uneven area.

However, when a color filter used for a color liquid crystal display device is manufactured by a photolithography method, dot unevenness having a diameter of about several pixels may occur in the color filter due to unavoidable factors in the manufacturing process. It is not possible at present to eliminate all. In addition, the diameter and shape of the unevenness, the density gradient at the boundary, and the like vary widely, and a digital filter of about 3 × 3 pixels, which has been frequently used in the related art, has a probability of successfully applying to such a shape of the dot-like unevenness. Therefore, it is difficult to detect dot-like unevenness.

[0004]

SUMMARY OF THE INVENTION The present invention has been made in view of such conventional problems, and an object of the present invention is to detect unevenness in quality inspection of a color filter or the like. An object of the present invention is to provide an image density unevenness detection method that improves detection sensitivity to unevenness and facilitates detection of dot-like unevenness.

[0005]

SUMMARY OF THE INVENTION The present invention analyzes a two-dimensional image obtained by photographing an object to be inspected, and detects an image density unevenness in which an area having a high or low density (an uneven area) is partially detected. 1) Coordinates (x ₀ , y) of a two-dimensional image obtained by shooting
₀ ) from the pixel located in the X-axis direction to Wx (Wx ≧ an integer)
1) An average luminance value (L _AP ) of all pixels located in an area (AP) occupied by Wy (Wy ≧ 1 as an integer) pixels in the Y-axis direction is produced. 2) A peripheral area of the area (AP) is produced. , From the area (AP) to D
(Integer D ≧ 1) Width H outside the pixel (H ≧ Integer
1) An average luminance value ( _LAS ) of all the pixels located in the donut-shaped area (AS) of the pixel is prepared. 3) The difference between the average luminance value ( _LAP ) and the average luminance value ( _LAS ) is calculated. This is a method of detecting unevenness in image density, wherein a non-uniform area is detected by comparing the threshold with a preset threshold value for unevenness detection.

According to the present invention, there is provided the image density unevenness detecting method according to the above invention, wherein the certain coordinates (x ₀ , y ₀ )
Is sequentially updated over the entire area of the two-dimensional image obtained by shooting, and an uneven density area is detected.

The present invention also relates to the image density unevenness detecting method according to the present invention, wherein a plurality of uneven areas are detected by a plurality of combinations of Wx and Wy in which Wx and Wy are changed. Is a method for detecting unevenness in image density, wherein the sum of the uneven areas is used as the uneven area of the inspection object.

[0008]

Embodiments of the present invention will be described below in detail. FIG. 1 to FIG. 4 are explanatory views schematically showing one embodiment of the image density unevenness detection method according to the present invention. FIG.
In (AT), pixels in the entire area of the two-dimensional image obtained by shooting are shown on XY orthogonal coordinates. A
(X ₀ , y ₀ ) indicates a pixel A located at a certain coordinate (x ₀ , y ₀ ) of the two-dimensional image.

FIG. 1 shows that Wx = 2 and Wy = 2, that is,
Coordinates (x ₀ , y ₀ ) of the two-dimensional image obtained by shooting
Is a case where all the pixels in the area (AP1) corresponding to the pixel A (A (x ₀ , y ₀ )) located in are 2 × 2 = 4 pixels. Also, D = 2, H = 1, that is, the donut-shaped area (AS1) has a width H = 1 pixel located outside the area (AP1) by D = 2 pixels. . Note that the area (AP1) is an area shown by oblique lines rising to the right, and the donut-shaped area (AS1) is an area shown by oblique lines falling to the right.

FIG. 2 shows that Wx = 5 and Wy = 5, that is,
Coordinates (x ₀ , y ₀ ) of the two-dimensional image obtained by shooting
Is a case where all the pixels of the area (AP2) corresponding to the pixel A (A (x ₀ , y ₀ )) located in the area A are 5 × 5 = 25 pixels. Also, D = 2, H = 1, that is, the donut-shaped area (AS2) is D = 2 from this area (AP2).
This shows a case where the width H = 1 pixel located outside the pixel. The area (AP2) is an oblique line rising to the right,
The donut-shaped area (AS2) is an area indicated by oblique lines falling to the right.

In this embodiment, Wx and Wy are 2 to 2.
An image density unevenness detection method that detects a plurality of uneven regions by a plurality of combinations of Wx and Wy changed in the range of 5, and uses the sum of the plurality of uneven regions as an uneven region of the inspection object explain. FIG. 3 shows one of a plurality of combinations of Wx and Wy when Wx and Wy are changed in the range of 2 to 5. That is, Wx = 4, Wy = 3, a pixel A located at a certain coordinate (x ₀ , y ₀ ) of a two-dimensional image obtained by shooting.
All pixels in the area (AP3) corresponding to (A (x ₀ , y ₀ )) are 4 × 3 = 12 pixels, and D = 2, H = 1, and the donut-shaped area (AS3) is AP3) to D =
This shows a case where the width H is one pixel located two pixels outside.

First, as shown in FIG. 1, D = 2 and H =
1 and Wx = 2, Wy = 2, and from a pixel A (A (x ₀ , y ₀ )) located at a certain coordinate (x ₀ , y ₀ ) of a two-dimensional image obtained by shooting. , Wx = 2 in the X-axis direction,
An average luminance value (L _AP1 ) of all pixels located in an area (AP1) occupied by Wy = 2 pixels in the Y-axis direction is obtained. L
_AP1 the (x _0, y _0), when the average luminance value of the pixel _{A (A (x 0, y} 0)) corresponding to Wx = 2, Wy = 2 regions _{(AP1), L AP1 (x} 0, y ₀ ) is represented by the following equation (1).

[0013] _{L AP1 (x 0, y 0} ) = ΣL (x i, y j) / (Wx × Wy) ··· (1) i = 0~ (Wx-1) j = 0~ (Wy-1 That is, in this case, (i = 0, j = 0), (i =
0, j = 1), (i = 1, j = 0), (i = 1, j =
This represents the average luminance value of the four pixels of 1).

Next, a width H = D = 2 pixels outside the area (AP1) around the area (AP1).
The average luminance value (L _AS1 ) of all pixels located in the donut-shaped area (AS1) of one pixel is obtained. L _AS1 (x ₀ , y ₀ )
Is the average luminance value of the donut-shaped area (AS1) corresponding to the pixel A (A (x ₀ , y ₀ )), L _AS1 (x ₀ ,
y ₀ ) is represented by the following equation (2). Incidentally, the donut-shaped area (AS1) has AS1a,
AS1b, AS1c, and AS1d divided into four parts were collected.

L _AS1 (x ₀ , y ₀ ) = {L _AS1a (x _i1 , y _j1 ) / (Wx + 2D + 2H)} + { _{LAS 1b} (x _i2 , y _j2 ) / (Wx + 2D + 2H)} + { _{LAS 1c} (x _{i3, y j3) / (Wy} + 2D)} + {ΣL AS1d (x i4, y j4) / (Wy + 2D)} ··········· (2) i 1 = - (D + H) ~ (( Wx-1) + D + H) j ₁ = − (D + H) i ₂ = − (D + H) to ((Wx−1) + D + H) j ₂ = (D + H) i ₃ = − (D + H) j ₃ = − (D) to ((Wy−1) + D) i ₄ = (D + H) j ₄ = − (D) to ((Wy−1) + D) L _AS1 (x ₀ , y ₀ ) = ｛ _{ΣLAS 1a} (x _i1 , y _j1 ) _{+ ΣL AS1b (x i2, y} j2) + ΣL AS1c (x i3, y j3) + ΣL AS1d (x i4, y j4)} / {2 (Wx + 2D + 2 H) +2 (Wy + 2D)} ··········・(3) That is, in this case, eight pixels (j = -3, i = -3 to 4), eight pixels (j = 3, i = -3 to 4), and (i = −
6 pixels of (3, j = −2 to 3), (i = 3, j = −2 to 3)
3) The average luminance value of a total of 28 pixels of 6 pixels is obtained. The expression (2) is obtained by dividing the donut-shaped area (AS1) into four parts, and the expression (3) is obtained by integrating the divided parts into four parts.

Subsequently, the average brightness value of the area (AP1), L _AP1 (x ₀ , y ₀ ), and the donut-shaped area (A
The difference between the average luminance value of S1) and L _AS1 (x ₀ , y ₀ ) is obtained. This difference is compared with a threshold value (S) for unevenness detection set in advance, and when the difference is larger than the threshold value, that is, when the relationship of the following equation (4) holds, the pixel A (A (x ₀ , y)
₀ )) is determined to be an uneven area. | L _AP1 (x ₀ , y ₀ ) −L _AS1 (x ₀ , y ₀ ) |> S (4) Then, the determination result is expressed as C (x ₀ , y ₀ ) = 1 [in the uneven area. There is]. However, before making a decision,
(X ₀ , y ₀ ) = 0 is set to [not a non-uniform area].

As described above, it is possible to determine whether or not a specific location in the inspection object is an uneven area. That is, it becomes possible by setting the coordinates of a certain specific place instead of the certain coordinates (x ₀ , y ₀ ). In the present invention, D = 2, H = 1, and Wx =
2, Wy = 2 is set in consideration of the diameter and shape of unevenness in the inspection object, the density gradient at the boundary, and the like. D is about 9 ≧ D ≧ 1, and H is 3 ≧ H. ≧ 1 and Wx are about 1 to 6, and Wx is about 1 to 6.

Further, the pixel A (A (x ₀ , y
₀ )), the above calculation is repeated for the entire two-dimensional image obtained by shooting while updating the coordinates one after another. By repeating this calculation, the setting condition (D = 2, H
= 1, and Wx = 2, Wy = 2) are detected for the entire two-dimensional image. (D + H) of the peripheral portion, upper side portion, and left side portion of the two-dimensional image
For pixels, (D + H + (Wx-1)) pixels on the right side, and (D + H + (Wy-1)) pixels on the lower side, the detection of dot-like unevenness is excluded.

Further, when D = 2 and H = 1, Wx and Wy
By detecting a plurality of uneven regions by a plurality of combinations of Wx and Wy in which is changed in the range of 2 to 5,
The accuracy of detecting the uneven area can be further improved. For example, from the settings shown in FIG. 1 (D = 2, H = 1, and Wx = 2, Wy = 2), one of the settings shown in FIG. 3 (D = 2, H = 1, and Wx = 4). , Wy =
3) through many combinations of Wx and Wy, such as 3)
(D = 2, H = 1, Wx = 5, Wy =
In a plurality of combinations up to 5), a plurality of uneven areas are detected. The final non-uniform area at this time is the non-uniform area of the inspection object with the sum of the plurality of non-uniform areas.

As described above, in the image density unevenness detection method according to the present invention, the area between the area of interest (AP) and the area to be compared (AS) is separated by D (an integer D ≧ 1) pixels outside. It is possible to avoid a difference in detection power depending on whether the density fluctuation at the edge of the uneven area is steep or gentle, and detect the uneven area with high sensitivity. In addition, by repeatedly detecting a plurality of uneven regions with a plurality of combinations of Wx and Wy while changing Wx and Wy, a difference in detection power due to uneven diameter, shape, density gradient at a boundary portion, and the like can be obtained. Thus, it is possible to detect the uneven area with high sensitivity.

FIG. 5 is an explanatory diagram showing an example of a flow of detecting an uneven area in the image density unevenness detecting method according to the present invention. First, the setting conditions in FIG. 1, D = 2, H = 1,
And the above C = 0 [not a non-uniform area], and Wx = 2, W
Set y = 2. Next, in FIG. 1, coordinates (x _a , x _a ) separated by (D + H + 1) pixels in the X-axis and Y-axis directions from the upper left end of the entire area of the captured two-dimensional image.
Set y _a ). Note that X _max and Y _max are each X
The data size in the axial direction and the data size in the Y-axis direction.

Next, Equation (1) a generalized equation _{(11) L AP (x,} y) = ΣL (x i, y j) / (Wx × Wy) ···· (11) i = 0- (Wx-1) j = 0 to (Wy-1) The coordinates (x
_a, a pixel B located to _{y a) (B (x a} , y a) from) of all pixels located in the region (AP) occupying the X-axis direction to Wx = 2, Y-axis direction of Wy = 2 pixels An average luminance value ( _LAP ) is obtained.

Next, an equation obtained by generalizing the equation (3)
(13) L_AS(X, y) = ｛ΣL_{AS a}(X_i1, Y_j1) + ΣL_{AS b}(X_i2, Y_j2) + ΣL _{AS c} (X_i3, Y_j3) + ΣL_{AS d}(X_i4, Y_j4) / {2 (Wx + 2D + 2H) +2 (Wy + 2D)} (13) According to (13), this area (A) at the periphery of this area (AP)
P) is a pixel having a width H = 1 pixel located outside D = 2 pixels from P).
The average luminance value (L) of all the pixels located in the nut-shaped area (AS)
_AS).

Subsequently, the average luminance value of the area (AP),
_{L AP (x a, y a} ) and the average luminance value of the donut-shaped region _{(AS), L AS (x} a, y a) the difference between the seek. This difference is compared with a preset threshold value for unevenness detection, and the threshold value (S)
When the relationship of Expression (14), which is a generalization of Expression (4) shown below, holds, the pixel B
_{(B (x a, y a} )) region corresponding to the (AP) is one in which it is determined that the uneven area. _{| L AP (x, y)} -L AS (x, y) |> S ···· (14) Then, the determination result and C (x, y) = 1 [ a uneven areas. Note that the judgment result before obtaining the difference is
C (x, y) = 0 (not a non-uniform area).

[0025] Subsequently, in the same manner x = x _{a + 1, i.e., L AP (x a + 1} , y a) of the pixels adjacent in the X-axis direction _{x a (x a + 1,} y a), L asked the _{AS (x a + 1, y} a),
The result of the difference determination is obtained. Then, sequentially x> ｛X _max −
(Get D + H + (Wx-1 ) determination result of each pixel in the X-axis direction up to the}. Then, y = y _{a + 1,} i.e., Y-axis of pixels _{B (B (x a, y} a)) Pixels adjacent in the direction (x
_a, y _{a + 1)} of _{L AP (x a, y a} + 1), L AS (x a, y
_{a + 1} ), and the result of the difference is obtained. Then, sequentially _{x> {X max - (D} + H + (Wx-1)} X up to the
The determination result of each pixel in the axial direction is obtained.

Similarly, sequentially, y> ｛Y _max- (D + H
+ (Wy-1)}, that is, the determination result up to all pixels in the X-axis direction of the pixel located at the lower left end of the two-dimensional image is obtained. That is, by the above, the setting condition (D
= 2, H = 1, and B = 0 [not a non-uniform area], and Wx = 2, Wy = 2), the result of determination of all pixels in the area except the peripheral part of the two-dimensional image (non-uniform area detection) ). Here, the peripheral portion includes (D + H) pixels on the upper side and the left side, and (D + H +) on the right side.
(Wx-1)) pixel, (D + H + (Wy-
1)) Pixels.

Next, the set conditions D = 2, H =
1, and C = 0 [not in the uneven area], and Wx =
2, Wy = 2, the values of Wx and Wy to be sequentially changed are set to 5, and the above processing is performed by a plurality of combinations of Wx and Wy in which Wx and Wy are changed. Is detected. The final non-uniform area is defined as the non-uniform area of the inspection object with the sum of the plurality of non-uniform areas.

[0028]

According to the present invention, a two-dimensional image obtained by photographing an object to be inspected is analyzed, and the image density unevenness detection for partially detecting a high or low density area (uneven area) includes: X from a pixel located at a certain coordinate (x ₀ , y ₀ )
Area occupied by Wx pixels in the axial direction and Wy pixels in the Y axis direction (A
An average luminance value (L _AP ) of all the pixels located in P) is created,
b) preparing an average luminance value ( _LAS ) of all pixels located in a donut-shaped area (AS) having a width of H pixels located outside the area (AP) by D pixels; and c) producing an average luminance value ( _LAP ). And the average luminance value (L _AS ) is compared with a preset threshold value of unevenness detection to detect unevenness of the image density. In the detection, the difference in the detection power depending on whether the density fluctuation at the edge of the uneven area is steep or gentle is avoided, and in particular, the detection sensitivity for point-like unevenness is improved,
An image density unevenness detection method that facilitates detection of dot-like unevenness is provided.

According to the present invention, in the image density unevenness detection method, a certain coordinate (x ₀ , y ₀ ) is sequentially updated over the entire area of a two-dimensional image obtained by photographing to detect an uneven area. Since it is an image density unevenness detection method, in unevenness detection in quality inspection such as color filters,
Avoids a difference in detection power depending on whether the density fluctuation at the periphery of the uneven area is steep or gentle over the entire area of the inspection object. An image density unevenness detection method that facilitates detection is provided.

Further, according to the present invention, in the above image density unevenness detecting method, a plurality of uneven areas are detected by a plurality of combinations of Wx and Wy with Wx and Wy changed, and the plurality of uneven areas are detected. Is a method for detecting unevenness in image density, which is a non-uniform area of the object to be inspected by adding the sum of the values. In particular, the present invention provides an image density unevenness detection method that improves the detection sensitivity for point-like unevenness and facilitates the detection of point-like unevenness.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram schematically showing one embodiment of a method for detecting unevenness in image density according to the present invention.

FIG. 2 is an explanatory diagram schematically showing one embodiment of a method for detecting unevenness in image density according to the present invention.

FIG. 3 is an explanatory diagram schematically showing an embodiment of a method for detecting unevenness in image density according to the present invention.

FIG. 4 is an explanatory view schematically showing an embodiment of the image density unevenness detection method according to the present invention.

FIG. 5 is an explanatory diagram showing an example of a flow of uneven area detection in the image density unevenness detecting method according to the present invention.

[Explanation of symbols]

_{A (x 0, y 0)} ... pixel A B (x _a, y _a) is located in a certain coordinate (x _0, y ₀₎ ... coordinate (x _a, y _a) pixel B AP ... Wx × Wy located Area occupied by pixels AP1 ... Wx = 2, Wy = 2 area corresponding to pixel A AP2 ... Wx = 5, Wy = 5 area corresponding to pixel A AP3 ... Wx = 4, Wy = 3 corresponding to pixel A Area AS... Donut-shaped area AS1... Wx = 2, Wy = 2, D =
2, H = 1 donut-shaped region AS2 ... Wx = 5, Wy = 5, D = corresponding to pixel A
2, H = 1 donut-shaped area AS3 ... Wx = 4, Wy = 3, D = corresponding to pixel A
2, H = 1 in the donut-shaped region AT ... shooting average luminance value of all pixels located in the pixel L _AP ... areas of all regions (AP) of the 2-dimensional images obtained by _{L AP1 (x 0, y 0} ) ... Wx = 2, W corresponding to pixel A
Average luminance value L _{AS of} y = 2 area (AP1)... Average luminance value of all pixels located in donut-shaped area (AS) L _AS1 (x ₀ , y ₀ ). Donut-shaped area (AS1) corresponding to pixel A ) Average luminance value of all pixels located in S): threshold value of unevenness detection X _max : data size in X-axis direction Y _max : data size in Y-axis direction

Claims (3)

[Claims] An image density unevenness detection for analyzing a two-dimensional image obtained by photographing an inspection object and partially detecting a high or low density area (uneven area) includes: Coordinates (x ₀ , y) of the obtained two-dimensional image
₀ ) from the pixel located in the X-axis direction to Wx (Wx ≧ an integer)
1) An average luminance value (L _AP ) of all pixels located in an area (AP) occupied by Wy (Wy ≧ 1 as an integer) pixels in the Y-axis direction is produced. 2) A peripheral area of the area (AP) is produced. , From the area (AP) to D
(Integer D ≧ 1) Width H outside the pixel (H ≧ Integer
1) An average luminance value ( _LAS ) of all the pixels located in the donut-shaped area (AS) of the pixel is prepared. 3) The difference between the average luminance value ( _LAP ) and the average luminance value ( _LAS ) is calculated. A method for detecting unevenness in the density of an image by comparing the threshold with a preset threshold value for unevenness detection. 2. The method according to claim 1, wherein the certain coordinates (x ₀ , y ₀ ) are sequentially updated over the entire area of the two-dimensional image obtained by the photographing, and an uneven area is detected. Image density unevenness detection method. 3. Wx and Wy obtained by changing Wx and Wy.
3. The image according to claim 1, wherein a plurality of non-uniform areas are detected by a plurality of combinations of the non-uniform areas, and the sum of the plurality of non-uniform areas is used as the non-uniform area of the inspection object. Shading unevenness detection method.