JP2707743B2 - Automatic evaluation method of degree of change of surface texture - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention uses an image analysis device to detect a wall or floor such as an outer wall.
The present invention relates to a method for evaluating the degree of change in surface properties such as dirt and scratches on a ceiling surface or various member surfaces, and peeling of paint.

[Prior art] As a method of quantifying the dirt on the floor and outer walls of a building,
Measure the reflectance, chromaticity, spectral reflectance, etc. of the exposed sample,
In the case of window glass, light transmittance is measured.

[Problems to be Solved by the Invention] The conventional dirt measurement method is based on comparison of physical properties such as reflectance and transmittance between a reference standard sample (new sample or unexposed sample) and an actual sample. It is for determining the degree, and a standard sample is always required as a criterion for the determination.

However, when judging the degree of dirt on the outer wall surface of an actual building, there is often no standard sample to be used as a reference, and control cannot be performed.

In addition, when a person looks at a member (for example, a wall surface or a floor) and determines whether or not the member is dirty,
Normally, the degree of contamination is not determined by comparison with a standard sample, but is determined based only on the surface properties of the sample visually, so that the conventional evaluation method based on comparison with the standard sample actually There is a possibility that the deviation from the dirt degree evaluation visually judged by a person may become large.

[Means for Solving the Problems] A method for automatically evaluating the degree of change in surface texture according to the present invention includes inputting an image of a surface to be evaluated to an image analyzer, and calculating the degree of change in surface texture based on the brightness value of each pixel. Related to the method of measuring.

In the present invention, a set of lightness values of each pixel is represented by Lt using a threshold Lt.
When divided into a group of lightness values smaller than Lt and a group of lightness values larger than Lt, two groups of variance σ _T σ _T = ωs (Ls−La) ² + ωb (Lb− La) ² ωs: ratio of pixels belonging to low lightness value group ωb: ratio of pixels belonging to high lightness value group Ls: average lightness of pixels belonging to low lightness value group Lb: average lightness of pixels belonging to high lightness value group La: all A threshold value Lt at which the average brightness of the pixel is maximized is determined, and the pixels of the low brightness value group and the high brightness value group pixels are determined based on the Lt value.

In the method of claim (1), the ratio of the number of pixels belonging to one of the determined groups to the total number of pixels is defined as Ar, and the average brightness difference between the determined two groups is defined as Ar.
When − ▲ is the contrast ΔL, the average brightness of the pixels belonging to the determined high brightness value group is
When Then, the degree of change of the evaluation target surface is evaluated by multiple regression.

According to the method of claim (2), the area Ss and the perimeter Ps of the low lightness value area formed by the pixels belonging to the low lightness value group determined as described above are obtained. And the ratio between the number of pixels belonging to one of the determined groups and the total number of pixels is defined as Ar, and the average brightness difference between the determined two groups is defined as the contrast ΔL. , Then, the degree of change of the evaluation target surface is evaluated by multiple regression.

According to the method of claim (3), the area Ss and the perimeter Ps of the low brightness value region formed by the pixels belonging to the low brightness value group determined as described above are obtained. Ask for. Then, when the ratio of the number of pixels belonging to one of the determined groups to the total number of pixels is defined as Ar, and the average brightness difference between the determined two groups is defined as the contrast ΔL, the determination is made. Average brightness of pixels belonging to the group of high brightness values The degree of change of the evaluation target surface is evaluated by multiple regression.

The method of claim (4) is: The value of It is treated as.

[Operation] The operation of the present invention will be described by taking, as an example, the case of evaluating the degree of contamination on the evaluation target surface according to the present invention.

When dirt is attached to the evaluation target surface, this dirt usually does not uniformly adhere to the entire evaluation target surface,
It adheres with some unevenness.

In the method of the present invention, whether each pixel belongs to a high brightness value group or a low brightness value group is classified based on the brightness value of each pixel of the image of the evaluation target surface.

In the present invention, a set of pixels belonging to the low brightness value group is treated as a contaminated portion. In the present invention, the threshold value for partitioning the high brightness value group and the low brightness value group is determined by the discriminant analysis method.

In order to extract features such as the area of the contaminated portion and the average brightness, the contaminated portion must be specified. However, the boundary of the contaminated portion is often not clear, and it is difficult to binarize such a grayscale image. Normally, in consideration of a situation where a person recognizes a contaminated portion, a boundary line where the difference between a dark portion and a light portion is clearest is assumed, and the former is recognized as a contaminated portion. In the present invention, a discriminant analysis method is applied as a binarization method that is close to human judgment. Specifically, in the brightness distribution of all the pixels taken in by the image analysis device, when the set of brightness values is divided into two groups by the threshold Lt, Lt is set such that the separation (variance) between the two groups is the largest. Determine the value. The pixel ratio and the average lightness of the low lightness value group (contaminated portion, L <Lt, where L is the lightness value of an arbitrary pixel) and the high lightness value group (the background portion with no or little contamination, L ≧ Lt) Are respectively ωs, ωb, Ls, Lb, and the overall average brightness is La, the variance σ _T between groups is ωs (Ls−La) ² + ωb (Lb−La)
^2, and it is Lt to do this maximum is determined. And
Based on the obtained Lt value that maximizes the variance, the division between the low lightness value (contaminated portion) group and the high lightness value group (non-contaminated portion) is determined.

The number of pixels belonging to one of the groups classified according to the determined threshold value (hereinafter, in the section of [Action], an example in which the low brightness value is set to the one group) will be described. Ar is the ratio of the total number of pixels captured and input.

In addition, the average brightness difference between the two groups confirmed ▲ ▼ −
▼ is the contrast ΔL. Further, the area of the contaminated region formed by the pixels belonging to the determined low lightness value group is Ss, and the circumference is Ps.

As a result of various studies conducted by the present inventor, the average brightness value of the contaminated part was ▲ ▼ and Was found to have a good correlation with the result of the visual judgment of the degree of contamination by the inspector. Claim (1) is based on this finding.

Similarly, as in claim (2), In the multiple regression with
▼, Also in the multiple regression, it was found that the judgment result of the visual contamination degree of the inspector and the multiple regression analysis had a good correlation.

[Example] Example 1 (Automatic evaluation of the degree of contamination of the mortar plate) As described below, the imaging of the surface of the mortar plate was image-analyzed, various physical quantities were measured, and the degree of contamination was automatically evaluated.

The measurement of the physical quantity was performed by an image analyzer (Nexus 6500) and a microcomputer. The color information of the sample (in this embodiment, a photograph of the sample) is obtained by a video camera using RG.
B (red, green, blue) is captured by the image analyzer as 256-level data. In this embodiment, the brightness information of the sample is targeted. First, from the gray scale measurement, the Y-stimulus value and RG
A correction equation with the B value was obtained (Equation-1). The Y-stimulus value was obtained from the G data of the sample input according to Equation 1, and was converted to L ^*, which is a lightness index (Equation 2).

Y = (0.875 + 0.0341 · G) ² Equation-1 L ^* = 116 (Y / Y ₀ ) ^1/3 −16 (Y ₀ = 100) Equation 2 Next, the variance σ defined by Equation 3 _The lightness threshold Lt that maximizes _T was determined.

σ _T = ωs (Ls−La) ² + ωb (Lb−La) ² Formula-3 where ωs, ωb, Ls, La, and Lb are as follows.

ωs: Ratio of pixels belonging to low lightness value group ωb: Ratio of pixels belonging to high lightness value group Ls: Average lightness of pixels belonging to low lightness value group Lb: Average lightness of pixels belonging to high lightness value group La: Average of all pixels Lightness The pixels of the low lightness value group and the pixels of the high lightness value group were determined using the threshold Lt as a boundary.

The ratio between the number of pixels belonging to the determined low lightness value group and the total number of pixels was defined as Ar, and the determined average lightness difference between the two groups was defined as contrast ΔL.

In addition, the K value of Expression-4 was binarized and then automatically measured to obtain a physical quantity representing the complexity of the contaminated portion shape.

FIGS. 1 and 2 show the results of determining various physical quantities by a discriminating method for a contaminated sample (contaminated mortar plate used in a sensory test described later). It can be said that the shape of the contaminated portion in the photograph and the binary image match well.

_{Incidentally, Lt, σ m, Ar,} Ls, Lb were as follows.

_{Lt = 45.3 σ m = 61.1 Ar} = 0.35 Ls = 37.3 Lb = 53.6 Then, the inspector for simulating contaminated sample was a sensory test (rank method) were measure the degree of contamination. This sample was a 14 cm × 16 cm mortar plate, which was contaminated by outdoor exposure. As the simulated contaminated samples, those having the patch-like contamination shown in FIGS. 3 (a) to 3 (i) attached (referred to as M series) and the flowing-down contamination shown in FIGS. 4 (a) to 4 (i) were used. Three types: those that adhered (referred to as S series) and those that adhered rectangular contamination (referred to as R series).
A series was used. FIGS. 3 to 5 schematically show the contamination, where a hatched area or a dot is a contaminated area, and the higher the density of the hatch or dot, the higher the degree of the contamination (the lower the brightness). ing.

Inspectors were 15 adults. The results of the sensory test are as shown in Table 1.

Tables 2 and 3 show the corresponding results when a single regression or a multiple regression analysis is performed on the physical quantity and the inspector's sensory quantity by the image analyzer. Table 2 shows the average lightness La, the variance σ _a ² of La, the average lightness Ls of the contaminated area, the variance σ _s ² of Ls, and the average lightness of the background.
▼, contaminated area ratio Ar, brightness difference ΔL, A · ΔL and 3 shows the results of a simple regression analysis of each physical quantity and the sensory quantity of the inspector. Table 3 shows the results of multiple regression analysis of combinations of two or three of the above physical quantities.

Example 2 (Evaluation of actual outer wall surface contamination) The inspector (35 adults) evaluated the degree of contamination of the outer wall surface of a reinforced concrete building (Utsunomiya University premises, achromatic finish). The evaluation was made into five stages and scaled by the sequential category method.

At the same time, a photograph of the same wall surface was taken, and the designated portion was taken into an image analyzer to determine a physical quantity. At that time, the color chart was simultaneously printed to correct the brightness.

Tables 4 and 5 show the analysis results. From Table-4 and 5, It is recognized that the multiple regression of the combination has a high multiple correlation coefficient.

It should be noted that although the average brightness La alone has a low correlation, it is presumed that this is due to the contamination of various shapes.

Has good correlation like the simulated sample. In multiple regression It can be said that the area of the contaminated portion, the contrast, the complexity coefficient, and the like are important factors in judging the degree of contamination since the combination of.

In the simple regression, those having a high correlation coefficient include the average lightness La, the contaminated portion lightness Ls, and the background lightness Lb. The simple regression does not show a good correlation between the contaminated area ratio Ar and the contrast ΔL, but the combination of both Ar · ΔL and its square root Has good correlation.

Even in multiple regression It can be said that the combination of ▼ and ▼ has a good correlation, and the criterion for determining the degree of contamination involves the area of the contaminated portion, the contrast, the brightness of the background, and the like.

[Effect] According to the method of the present invention, as is clear from the above embodiments, it is possible to determine the degree of contamination of the evaluation target surface only by performing image processing on the image of the evaluation target surface without comparing with the standard sample. Can be. This determination is made comprehensively based on the size, shape, color density, and the like of the contaminated portion, and a result very similar to the result of human visual determination is obtained.

According to the method of the present invention, not only contamination of various members such as a wall surface but also visual appearance of a portion which can be binarized by a threshold value such as paint peeling or cracking can be visually judged.

[Brief description of the drawings]

1 and 2 are distribution diagrams showing analysis results, and FIGS. 3, 4, and 5 are schematic diagrams of a sample to be inspected.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Kazumasa Goto 3-6, Koiehonmachi, Tokoname-shi, Aichi Prefecture Inax Corporation (72) Inventor Kazuhiro Kishira 3-6, Koiehonmachi, Tokoname-city, Aichi Prefecture (72) Inventor Yoshinori Tachibana 950-13 Shimohiraide-cho, Utsunomiya, Tochigi

Claims (4)

(57) [Claims] 1. A method of inputting an image of an evaluation target surface to an image analysis device and measuring a degree of change in surface texture based on a brightness value of each pixel, wherein a set of brightness values of each pixel is determined by a threshold Lt. When divided into a group of lightness values smaller than Lt and a group of lightness values larger than Lt, two groups of variance σ _T σ _T = ωs (Ls−La) ² + ωb (Lb −La) ² ωs: ratio of pixels belonging to low lightness value group ωb: ratio of pixels belonging to high lightness value group Ls: average lightness of pixels belonging to low lightness value group Lb: average lightness of pixels belonging to high lightness value group La: A threshold Lt at which the average brightness of all pixels is maximized is determined, and the pixels of the low brightness value group and the pixels of the high brightness value group are determined based on the Lt value, and belong to one of the determined groups. Ar is the ratio of the number of pixels to the total number of pixels, and Average lightness difference of both groups ▲ ▼ - ▲ ▼ contrast Δ
L, the average brightness of the pixels belonging to the determined high brightness value group An automatic evaluation method for the degree of change in surface texture, wherein the degree of change in the surface to be evaluated is evaluated by multiple regression with the above. 2. A method of inputting an image of a surface to be evaluated to an image analyzer and measuring a degree of change in surface texture based on a brightness value of each pixel, wherein a set of brightness values of each pixel is determined by a threshold Lt. When divided into a group of lightness values smaller than Lt and a group of lightness values larger than Lt, two groups of variance σ _T σ _T = ωs (Ls−La) ² + ωb (Lb −La) ² ωs: ratio of pixels belonging to low lightness value group ωb: ratio of pixels belonging to high lightness value group Ls: average lightness of pixels belonging to low lightness value group Lb: average lightness of pixels belonging to high lightness value group La: A threshold Lt at which the average brightness of all the pixels is maximized is determined, and the pixels of the low brightness value group and the high brightness value group are determined based on the Lt value, and the pixels belonging to the determined low brightness value group are determined. Area Ss and perimeter Ps of the low brightness value area where La And the ratio of the number of pixels belonging to one of the determined groups to the total number of pixels is defined as Ar, and the determined average brightness difference between the two groups is defined as the contrast ΔL.
And when An automatic evaluation method for the degree of change in surface texture, wherein the degree of change in the surface to be evaluated is evaluated by multiple regression with the above. 3. A method of inputting an image of a surface to be evaluated to an image analysis device and measuring a degree of change in surface texture based on a brightness value of each pixel, wherein a set of brightness values of each pixel is determined by a threshold Lt. When divided into a group of lightness values smaller than Lt and a group of lightness values larger than Lt, two groups of variance σ _T σ _T = ωs (Ls−La) ² + ωb (Lb −La) ² ωs: Pixel ratio belonging to low lightness value group ωb: Pixel ratio belonging to high lightness value group Ls: Average lightness of pixels belonging to low lightness value group Lb: Average lightness of pixels belonging to high lightness value group La: A threshold Lt at which the average brightness of all the pixels is maximized is determined, and the pixels of the low brightness value group and the high brightness value group are determined based on the Lt value, and the pixels belonging to the determined low brightness value group are determined. And the area Ss and perimeter Ps of the low brightness value area Et al. When the ratio between the number of pixels belonging to one of the determined groups and the total number of pixels is defined as Ar, and the average brightness difference between the determined two groups is defined as the contrast ΔL, Average brightness of pixels belonging to the group of the determined high brightness value ▲ ▼, An automatic evaluation method for the degree of change in surface texture, wherein the degree of change in the surface to be evaluated is evaluated by multiple regression. 4. When performing multiple regression, The value of The method according to claim 2 or 3, wherein the method is treated as: