JP2014109799A - Pattern position detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an accurate and fast pattern position detection method.SOLUTION: A pattern position detection method includes: a step (S101) for creating a reference image; a step (S102) for acquiring a captured image in which a position including a pattern part on a target medium is photographed; and a step (S103) for performing pattern position detection based on pattern matching using the reference image created in the step S101 and the captured image acquired in the step S102.

Description

  The present invention relates to a pattern position detection method, and more particularly to a pattern position detection method based on pattern matching.

  When producing clothing and other products, the fabric is finished through a process of cutting fabrics, knitted fabrics, non-woven fabrics, etc. to create the original parts of the product, and joining the created parts by sewing etc. It is common. Furthermore, when a pattern is provided in advance on the fabric before cutting, a part is created by cutting according to the pattern provided in general. Conventionally, the pattern provided to the fabric has been created by a dyeing process. However, in recent years, it has also been proposed to create a pattern by using an ink jet recording apparatus.

  As means for cutting a fabric on which a pattern is created, cutting by an automatic cutting machine is known in addition to conventional manual cutting. When cutting with an automatic cutting machine, the cutting is usually performed according to predetermined cutting data set by an image processing apparatus such as CAD. When actually cutting a fabric with a patterned pattern, the fabric tends to be distorted when the fabric is set on the cutting stage, and the patterned pattern of the fabric is different from the position of the patterned pattern corresponding to the set cutting data. May be granted. In this case, it is necessary to correct the cutting position according to the actual pattern position.

  However, in the case of the cutting method for recognizing the pattern pattern applied to the fabric and correcting the cutting position, the pattern pattern bleeds (hereinafter referred to as noise) due to the surface state of the fabric and the diffusion state of the coloring material. There was a problem that the cutting position could not be corrected.

  In order to cope with this, for example, Patent Document 1 describes a method of identifying noise having high density in image data and removing the noise without blurring the edge of the image. However, although this method can reduce noise in a captured image, it has been extremely difficult to completely remove noise in the captured image.

JP 04-316275 A

  The present invention has been made in view of such a current situation, and an object of the present invention is to provide a pattern position detection method that can be performed with high accuracy and high speed.

The present invention
Creating a reference image (S101);
A step (S102) of acquiring a captured image obtained by photographing a position including a pattern portion on the target medium;
This is a pattern position detection method including a step (S103) of detecting a pattern position by pattern matching using the reference image created in step S101 and the captured image acquired in step S102.

  By using the method of the present invention, it is possible to detect a pattern position by high-precision and high-speed pattern matching for a captured image of a target medium in which noise has occurred.

Process Flow for Pattern Pattern Position Detection Method of the Present Invention Design image Target media Target medium after design pattern has been added Model image in diffusion state ((a): t = 0, (b): t = 1, (c): t = 2) Reference image Captured image Pattern position detection result by pattern matching using reference image Result of pattern position detection by pattern matching using design image

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. The embodiments described below are preferable specific examples for carrying out the present invention, and thus various technical limitations are made. However, the present invention is particularly limited in the following description. Unless otherwise specified, the present invention is not limited to these forms.

  Hereinafter, a fabric is taken up as a target medium to which a design pattern having a design image as original data is given, and the design pattern position detection method of the present invention will be described according to the processing flow of FIG.

[Create Reference Image (S101)]
Depending on the surface state of the fabric serving as the target medium and the diffusion state of the color material, noise due to bleeding may occur in the applied pattern. For example, the pattern of the design image shown in FIG. 2 is changed to the target medium shown in FIG. When it is applied to the cloth, as shown in FIG. 4, noise due to diffusion of the color material is generated in the applied pattern. Therefore, in step S101, noise generated in the pattern applied to the fabric is predicted, and a reference image is created by applying equivalent noise to the design image. As a method of predicting noise, for example, noise prediction can be performed by obtaining a color material diffusion state based on Fick's second law expressed by the following equation (1).

Here, φ is the color material density, t is time, x and y are two-dimensional coordinates in the image, and D is a diffusion coefficient.

  When the equation (1) is discretized, the following equation (2) is obtained.

  According to Expression (2), the pixel value of the pixel at the position i, j at time t + 1 can be calculated. By using Expression (2) for each pixel of the design image, the pixel value of each pixel and the diffusion to the adjacent pixels can be calculated. Since the diffusion coefficient D can be obtained from experimental values for each target fabric, the diffusion coefficient D of the target medium may be measured in advance. If the diffusion coefficient D is determined, the noise generated in the pattern applied to the fabric can be predicted by calculating according to the expression (2) for each pixel of the design image.

  Here, the state of diffusion according to Equation (2) will be described with reference to the model image shown in FIG. When the initial state t = 0, it is assumed that only the central pixel has a pixel value ((a) in FIG. 5). When the diffusion coefficient D> 0 and the time is advanced to t = 1 according to the equation (2), diffusion occurs in the adjacent pixel of the center cell ((b) in FIG. 5). It diffuses to the pixels ((c) of FIG. 5). This blur due to diffusion becomes noise. An image obtained by adding noise to the design image according to Expression (2) is set as a reference image (FIG. 6) used in the subsequent step S103.

[Acquired Captured Image of Fabric (S102)]
In step S102, a captured image is acquired from the fabric provided with the design image. An example of the captured image is shown in FIG.

[Detection of pattern position by pattern matching between reference image and captured image (S103)]
In step S103, pattern position detection by pattern matching is performed using the reference image (FIG. 6) created in step S101 and the captured image (FIG. 7) acquired in step S102. A known method may be used as a pattern matching method, and is not particularly limited. For example, a template matching method using a reference image as a template image can be used.

  A template matching result between the reference image created in step S101 and the captured image acquired in step S102 is shown in FIG. 8, and the detected pattern position is indicated by a rectangle 1 in FIG. For accuracy comparison with the present invention, a template matching result when a design image is used as a template image is shown in FIG. 9, and the detected pattern position is indicated by a rectangle 2 in FIG. Comparing each detection position, the pattern pattern is correctly recognized in the template matching using the reference image, whereas in the template matching using the design image, the detection position is several pixels lower right than the pattern pattern. It can be seen that there is a shift (the part surrounded by a circle in FIG. 9). When the noise generated in the fabric is large, this deviation is further increased. For example, in a cutting method for recognizing a pattern and correcting a cutting position, a deviation of several pixels leads to a reduction in cutting quality.

  In this way, the noise of the target medium is predicted, a reference image is created by adding the predicted noise to the design image, and the obtained reference image is used for pattern matching. It can be performed at high speed.

Claims (1)

Creating a reference image (S101);
A step (S102) of acquiring a captured image obtained by photographing a position including a pattern portion on the target medium;
Using the reference image created in step S101 and the captured image obtained in step S102, including a step (S103) of detecting a pattern position by pattern matching.
Pattern pattern position detection method.