Blemish evaluation method, droplet evaluation method, repellent evaluation method, and repellent evaluation device
An evaluation method for evaluating a spot generated on a substrate that has been treated with a repellent, the method includes an acquiring step of acquiring image data including data on the substrate that has been treated with the repellent to be evaluated, a spot-detecting step of creating a smoothed image from the image data, binarizing the smoothed image and detecting a region of a spot generated on the substrate, and a spot-evaluating step of determining an evaluation value of the spot according to an area of the region of the spot detected. Also disclosed is a repellent evaluation device including an acquire, a classifier, a detector and an evaluator.
1 . An evaluation method for evaluating a spot generated on a substrate that has been treated with a repellent, the method comprising:
an acquiring step of acquiring image data including data on the substrate that has been treated with the repellent to be evaluated;
a spot-detecting step of (1) creating a smoothed image from a grayscale image of the image data, (2) obtaining an image by, for each image pixel in the grayscale image, subtracting a brightness value of the smoothed image from a brightness value of the grayscale image and adding an average of the brightness value of the gray scale image, (3) binarizing the obtained image, and detecting a region of a spot generated on the substrate; and
a spot-evaluating step of determining an evaluation value of the spot according to an area of the region of the spot detected.
2 . A repellent evaluation method, the method being for evaluating performance of the repellent and comprising:
an acquiring step of acquiring image data including data on a substrate that has been treated with the repellent to be evaluated;
a detecting step of detecting at least one of a region of a spot or a region of a water drop on the substrate in the image data; and
an evaluating step of determining an evaluation value of the performance of the repellent according to an area of the region of the spot detected or the number of the regions of the water drops detected,
wherein the detecting step for detecting the region of the spot, (1) creating a smoothed image from a grayscale image of the image data, (2) obtaining an image by, for each image pixel in the grayscale image, subtracting a brightness value of the smoothed image from a brightness value of the grayscale image and adding an average of the brightness value of the grayscale image, (3) binarizing the obtained image, and detecting a region of a spot generated on the substrate, and
wherein the detecting step for detecting the region of the water drop, creating an edge image based on whether or not the difference between the brightness values of adjacent image pixels of the image data is the predetermined value or more, and detecting the region of the water drop on the substrate based on the edge image.
3 . The repellent evaluation method according to claim 2 , wherein in the detecting step,
when the substrate includes a spot, the region of the spot is detected from the image data that has applied a first image processing for spot image data, and
when the substrate includes no spot, the region of a water drop is detected from the image data that has applied a second image processing for water drop image data.
4 . The repellent evaluation method according to claim 3 , wherein in the detecting step, a smoothed image data is created from the image data by the first image processing.
5 . The repellent evaluation method according to claim 3 , wherein in the detecting step, an edged image data is created from the image data by the second image processing.
6 . The repellent evaluation method according to claim 2 , further comprising a classifying step of classifying the image data as a spot image data when there are no wrinkle on the substrate in the image data and includes one or more spots, and classifying the image data as a water drop image data when the substrate includes a water drop with no spot.
7 . The repellent evaluation method according to claim 6 , wherein the classifying step achieves classifying the image data into a first spot image data and a second spot image data according to a spot-including pattern, and
in the detecting step, the region of a spot is detected by performing image processing different between a case where the image data is classified as the first spot image data and a case where the image data is classified as the second spot image data.
8 . The repellent evaluation method according to claim 7 , wherein in the detecting step, image processing for decreasing a tone is performed on the first spot image data, followed by performing binarization to detect the region of a spot, and image processing for increasing a contrast is performed on the second spot image data, followed by performing binarization to detect the region of a spot.
9 . The repellent evaluation method according to claim 6 , wherein in the classifying step, the image data acquired is used as input data, and the input data is classified by utilizing a model that has learned, by machine learning, a relation between a plurality of training image data including data on a substrate that has been treated with a repellent and a classification of the training image data.
10 . A repellent evaluation device for evaluating performance of the repellent and comprising:
at least one of a processor or hardware circuitry configured to implement;
an acquirer that acquires image data including data on a substrate that has been treated with the repellent to be evaluated,
a classifier that classifies the image data as spot image data when the substrate includes a spot in the image data and classifies the image data as water drop image data when the substrate includes no spot in the image data,
a detector, wherein when the image data is spot image data, the detector (1) creates a smoothed image from a grayscale image of the image data, (2) obtains an image by, for each image pixel in the image, subtracting a brightness value of the smoothed image from a brightness value of the gray scale image and adding an average of the brightness value of the gray scale image, (3) binarizes the obtained image, and detecting a region of a spot generated on the substrate, and when the image data is water drop image data, the detector creates an edge image based on whether or not the difference between the brightness values of adjacent image pixels of the image data is the predetermined value or more, and detects the region of the water drop on the substrate based on the edge image, and
an evaluator that determines an evaluation value of the performance of the repellent according to an area of the region of a spot detected or the number of the regions of the water drops detected.
11 . The repellent evaluation device according to claim 10 , wherein the classifier uses the image data acquired by the acquirer as input data, and classifies the input data by utilizing a model that has learned, by machine learning, a relation between a plurality of training image data including data on a substrate that has been treated with the repellent and a classification of the training image data.