KR101215569B1 - Method of segmenting for label region - Google Patents

Abstract

PURPOSE: A label area division method is provided to enable a user to receive or purchase a product including a label by recognizing a label area from an input image. CONSTITUTION: An importance area creation unit creates a predetermined area of an input image as an importance area(S1000). A vertical edge detection unit detects two vertical edges from the importance area and creates an area which is covered with the vertical edges as a label candidate area(S2000). A label area encoding unit creates encoded label area code by calculating the similarity of a direction and a size between pixels in a boundary of the label candidate area(S4000). A label area separation unit separates an area matched with the label area code from the input image(S5000). [Reference numerals] (S1000) Creating an importance area; (S2000) Creating vertical edges and a label candidate area which is covered by the vertical edges; (S3000) Detecting horizontal edges which cross with vertical edges; (S4000) Creating a label area code connected to vertical edges and horizontal edges; (S5000) Dividing a label area matching with a label code area in an inputted image

Description

Method of Segmenting for label region}

The present invention relates to a label region dividing method, and more particularly, to a label region dividing method capable of automatically recognizing and dividing a label region, which is an area where the label is located, without a user manually operating a label such as a bottle or canned food in an image. It is about.

In general, the image recognition field is a field for recognizing a specific object from an input image or an input image input from a photographing apparatus including a camera. Recently, a mobile terminal including a smart phone is commercially available and the camera is connected to the mobile terminal. As is basically provided, research and development for the user to recognize a particular object easily and conveniently using the mobile terminal is in progress.

In addition, in order to recognize the specific object, the object to be recognized in the input image or the input image should be divided. In this case, Gaussian mixture model (GMM), Markov random field (MRF), CV-Model, and K-average algorithm The object is partitioned using various algorithms such as k-means or importance map.

In addition, by using the mobile terminal equipped with the camera it is possible to recognize the object including the 'text', 'fingerprint' or 'poster' in the input image or the input image.

On the other hand, when the object to be recognized is a label attached to a bottle or canned food, the color and shape are complicated, so that the target label to be recognized in the input image or the input image cannot be accurately divided and the recognition is deteriorated. There was a problem.

In addition, in order to accurately divide the target object such as the label, the algorithms require the user to draw the outline of the target object directly, which causes a problem of time consuming and troublesome for the user. .

The present inventors have made efforts to automatically divide a label attached to a bottle or canned food in an input image to recognize the label as a label area. As a result, the present inventors have developed a technical configuration of a label area division method.

Accordingly, it is an object of the present invention to provide a label region segmentation method that can automatically segment a label region in an input image without the user directly manipulating it.

The objects of the present invention are not limited to the above-mentioned objects, and other objects not mentioned can be clearly understood by those skilled in the art from the following description.

In order to achieve the above object, the present invention provides a label region segmentation method for dividing the label region from the input image so that the computer apparatus can recognize a label in the input image. A first step of functioning to generate a predetermined region of the importance region; A second step of vertical edge detecting means for causing the computer device to detect two vertical edges in the region of importance and to generate an area surrounded by the vertical edges as a label candidate region; A third step of causing label area encoding means to generate a label area code in which the computer device calculates and encodes a similarity of directions and sizes between pixels at boundaries of the label candidate area; And a fourth step of the label region dividing means for causing the computer device to divide the region matching the label region code in the input image into the label region.

In another aspect, the present invention provides a computer-readable storage medium storing a label region segmentation program for performing the label region segmentation method by functioning as a means of any one of claims 1 to 9.

In a preferred embodiment, between the second step and the third step, a horizontal edge detecting means functions to cause the computer device to detect two horizontal edges crossing the vertical edges in the label candidate area. Step A; further comprises.

In a preferred embodiment, the first step may include: steps 1-1, wherein the importance region generating means functions to cause the computer device to convert a format of the input image into a lab color space; And calculating a region consisting of pixels whose directionality, color, and brightness are greater than or equal to a threshold value among the pixels of the converted input image to generate the region of importance.

In a preferred embodiment, the vertical edge detecting means is a step 2-1 wherein the computer device functions to generate an edge map of the region of importance using a sobel edge detector; And a step 2-2 of detecting a plurality of vertical edges of the straight edges after detecting a plurality of straight edges in the edge map.

In a preferred embodiment, the horizontal edge detecting means in step A functions to cause the computer device to detect the horizontal edges by calculating a structural tensor of the label candidate region.

In the third embodiment, the label area encoding means calculates a similarity in direction and size between pixels in the boundary line of the label candidate area using an eight-way chain code, and calculates the horizontal edges and the horizontal edges. It functions to generate a label area code with vertical edges connected.

In a preferred embodiment, the importance region generating means functions to cause the computer apparatus to generate the importance region using Equation 1 below.

[Equation 1]

는 중요도 영역이며,

는 방향 특징맵이고,

는 색상 특징맵이고,

은 밝기 특징맵이며,

는 각 특징맵의 결합을 위한 가중이고,

는 각 특징맵의 분산이다.here,

Is the importance zone,

Is the direction feature map,

Is the color feature map,

Is the brightness feature map,

Is the weight for combining each feature map,

Is the variance of each feature map.

In a preferred embodiment, the vertical edge detecting means functions to cause the computer device to detect the vertical edges using Equation 2 below.

&Quot; (2) "

는 원점에서 직선성분이 교차하는 점의 좌표값까지의 수직거리로써 세로 에지를 이루는 직선을 나타내며,

는 상수로써 x-축에 대해 정규(normal) 방향으로 이루어진다.here,

Is the vertical distance from the origin to the coordinate value of the point where the linear components intersect.

Is a constant and is made in the normal direction with respect to the x-axis.

In the preferred embodiment, the transverse edge detecting means functions to calculate the structural tensor and detect the transverse edges by using Equation 3 below.

&Quot; (3) "

는 구조 텐서를 뜻하고,

는 입력이미지를 뜻하고, x 및 y는 변수이고,

는 x의 도함수이며,

는 y의 도함수이다.here,

Stands for structural tensor,

Is the input image, x and y are variables,

Is the derivative of x,

Is the derivative of y.

The present invention has the following excellent effects.

First, according to the label region dividing method according to an embodiment of the present invention,

The effect can be obtained.

1 is a block diagram illustrating a label region dividing method according to an embodiment of the present invention.

Although the terms used in the present invention have been selected as general terms that are widely used at present, there are some terms selected arbitrarily by the applicant in a specific case. In this case, the meaning described or used in the detailed description part of the invention The meaning must be grasped.

Hereinafter, the technical structure of the present invention will be described in detail with reference to preferred embodiments shown in the accompanying drawings.

However, the present invention is not limited to the embodiments described herein but may be embodied in other forms. Like reference numerals designate like elements throughout the specification.

The label region segmentation method according to an embodiment of the present invention is to divide only the label region from the input image and remove noise other than the label region so that the computer apparatus can recognize the label region in the input image. The attached product can be used to receive or purchase information.

In addition, the label area dividing method according to an embodiment of the present invention is substantially stored in the form of a program in a mobile terminal, PDA terminal or tablet PC equipped with a computer or a computing processor to perform a function.

In addition, the program is stored in a computer readable medium to perform the function by being read by a computer or mobile terminal, PDA terminal or tablet PC, the medium is those specially designed and configured for the present invention or computer software It may be known and available to those of ordinary skill in the art, for example, magnetic media such as hard disks, floppy disks and magnetic tape, optical recording media such as CD, DVD, magnetic and optical recording A magnetic device, specially configured to store and execute the program instructions, alone or in combination, such as a magnetic-optical recording medium, ROM, RAM, flash memory, and the like.

In addition, the program is not only read by a computer, a mobile terminal, a PDA terminal or a tablet PC by the medium, but also stored in a server system capable of transmitting information through a communication network such as an intranet or the Internet, and then transmitted to a computer. Alternatively, the server system may provide a platform on which the computer may access the server system and execute the program on the server system without transmitting the program to the computer.

Hereinafter, a method of dividing a label region according to an embodiment of the present invention will be described in detail.

1 is a block diagram illustrating a method of dividing a label region according to an embodiment of the present invention.

Referring to FIG. 1, in the label area segmentation method according to an embodiment of the present invention, first, when an input image photographed by a camera or a digital camera is input to a computer device, the importance region generating means may be configured by the computer device. It functions to generate a predetermined area of the area of importance (S1000).

The computer apparatus may also include a mobile terminal, a PDA terminal, or a tablet PC equipped with a computer or a computing processor, and the importance region generating means is a program or a program functioning to cause the computer apparatus to generate the importance region in the input image. It may be a means provided in.

In addition, the importance region generating means is provided so that the computer apparatus generates the importance region using a importance map algorithm.

In addition, the importance region generating means causes the computer device to convert the format of the input image, wherein the format of the input image is converted into a Lab color space.

The importance region generating means may be provided to adjust the size of the input image to a preset reference value.

In addition, the importance region generating means causes the computer apparatus to calculate the importance region, wherein the importance region generating means uses the importance map algorithm to determine the angle of the input image converted by the computer apparatus into the Lab color space. A region consisting of pixels whose directionality, color, and brightness are greater than or equal to a threshold value among pixels is calculated and generated as the importance region.

The importance region generating means calculates a direction feature map, a color feature map, and a brightness feature map from the input image and combines the feature maps to generate the importance region.

In addition, the importance region generating means may function to generate the importance region by using Equation 1 below.

[Equation 1]

는 중요도 영역이며,

는 방향 특징맵이고,

는 색상 특징맵이고,

은 밝기 특징맵이며,

는 각 특징맵의 결합을 위한 가중이고,

는 각 특징맵의 분산이다.here,

Is the importance zone,

Is the direction feature map,

Is the color feature map,

Is the brightness feature map,

Is the weight for combining each feature map,

Is the variance of each feature map.

In other words, the importance map generating means enables the computer device to generate the importance region in the most prominent portion of the input image.

Next, the vertical edge detecting means causes the computer apparatus to generate a label candidate region in the importance region, wherein the label candidate region detects a plurality of vertical edges in the importance region and covers the region surrounded by the vertical edges. The candidate area is generated (S2000).

The longitudinal edge detecting means also uses a sobel edge detector to cause the computer device to generate an edge map of the region of importance. Also, in the edge map, outlines of objects in the importance region are clearly displayed.

The longitudinal edge detecting means also allows the computer apparatus to detect a plurality of straight edges in the edge map using a Hough Transform algorithm. In addition, the longitudinal edge detecting means causes the computer device to detect two longitudinal edges of high straightness and longitudinal direction among the straight edges.

In addition, the vertical edges may be symmetrical to each other or a straight edge having the most similar brightness value, and the vertical edge detecting means may obtain the slope of the vertical edges.

In addition, the vertical edge detecting means may function to cause the computer device to detect the vertical edges using Equation 2 below.

&Quot; (2) "

는 원점에서 직선성분이 교차하는 점의 좌표값까지의 수직거리로써 세로 에지를 이루는 직선을 나타내며,

는 상수로써 x-축에 대해 정규(normal) 방향으로 이루어진다.here,

Is the vertical distance from the origin to the coordinate value of the point where the linear components intersect.

Is a constant and is made in the normal direction with respect to the x-axis.

In addition, the vertical edge detecting means causes the computer device to label the area surrounded by the vertical edges as the label candidate area.

In addition, the vertical edge detecting means may be a program or a means provided to the computer device to detect the vertical edges in the importance region and to generate the label candidate region.

Next, horizontal edge detecting means functions to detect the two horizontal edges in which the computer apparatus intersects the vertical edges in the label candidate region (S3000).

In addition, the horizontal edge detecting means detects the horizontal edges by computing a structural tensor of the label candidate region and singular value decomposition of the structural tensor. In addition, the horizontal edges may be straight, but may also be curved or curved.

In addition, the horizontal edge detecting means functions to allow the computer device to calculate the structural tensor, detect the horizontal edges, and calculate the degree of inclination of each horizontal edge using Equation 3 below.

&Quot; (3) "

는 구조 텐서를 뜻하고,

는 입력이미지를 뜻하고, x 및 y는 변수이고,

는 x의 도함수이며,

는 y의 도함수이다.here,

Stands for structural tensor,

Is the input image, x and y are variables,

Is the derivative of x,

Is the derivative of y.

In addition, the horizontal edge detecting means may be a program or a means provided in the program that the computer device functions to detect the horizontal edges using the structural tensor.

Next, label area encoding means causes the computer device to generate a label area code obtained by calculating a degree of similarity between directions and sizes between pixels at the boundary of the label candidate area (S4000).

In addition, the label area encoding means may perform encoding by calculating a degree of similarity in direction and size along the pixels in which the vertical edges and the horizontal edges are positioned.

In addition, the label area encoding means uses an 8 directionnal chain code to calculate the similarity in the direction and size between the pixels by the computer device, and the label area code to which the horizontal edges and the vertical edges are connected. Function to generate

Further, the label area encoding means is provided to generate the label area code while calculating the similarity by comparing the directions and sizes of pixels in the 3x3 mask clockwise from the center of the left lateral line of the image.

The label area encoding means may be a program or a means provided in the program functioning to cause the computer device to generate the label area code using the eight-way chain code.

Label area dividing means then functions to cause the computer device to divide an area in the input image that matches the label area code into the label area, wherein the label area is substantially positioned in the input image. It may be a recognized area (S5000).

That is, the computer device may recognize a specific character or picture in the label area and display it to the user or provide another service.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation in the present invention. Various changes and modifications will be possible.

Claims (10)

A label region segmentation method for segmenting the label region in the input image so that a computer device can recognize a label in the input image,
A first step of generating, by the importance region generating means, the computer device to generate a predetermined region of the input image as an importance region;
A second step of vertical edge detecting means for causing the computer device to detect two vertical edges in the region of importance and to generate an area surrounded by the vertical edges as a label candidate region;
A third step of causing label area encoding means to generate a label area code in which the computer device calculates and encodes a similarity of directions and sizes between pixels at boundaries of the label candidate area; And
And a fourth step of the label area dividing means for causing the computer device to divide the area matching the label area code in the input image into the label area.
The method of claim 1,
Between the second step and the third step,
And a horizontal edge detecting means for causing the computer device to detect two horizontal edges intersecting the vertical edges in the label candidate region.
The method of claim 1,
In the first step,
The importance region generating means is the computer device,
A first step of converting a format of the input image into a lab color space; And
A first and second steps of calculating a region consisting of pixels whose orientation, color, and brightness are greater than or equal to a threshold among the pixels of the converted input image; .
The method of claim 1,
The vertical edge detecting means is the computer device,
A second step of functioning to generate an edge map of the region of importance using a sobel edge detector; And
And detecting a plurality of straight edges in the edge map, and then performing a second step of detecting two vertical edges of the straight edges. 2.
The method of claim 2,
And wherein in the step A, the horizontal edge detecting means functions to detect the horizontal edges by calculating a structural tensor of the label candidate region.
The method of claim 2,
In the third step, the label area encoding means calculates a similarity in direction and size between pixels in the boundary line of the label candidate area by using an eight-way chain code, and the label area to which the horizontal edges and the vertical edges are connected. And a label area segmentation function.
The method of claim 3,
And the importance region generating means functions to cause the computer apparatus to generate the importance region using Equation 1 below.
[Equation 1]

here,

Is the importance zone,

Is the direction feature map,

Is the color feature map,

Is the brightness feature map,

Is the weight for combining each feature map,

Is the variance of each feature map.
5. The method of claim 4,
And the vertical edge detecting means functions to detect the vertical edges by using the following Equation (2).
&Quot; (2) "

here,

Is the vertical distance from the origin to the coordinate value of the point where the linear components intersect.

Is a constant and is made in the normal direction with respect to the x-axis.
6. The method of claim 5,
And the horizontal edge detecting means functions to calculate the structural tensor and detect the horizontal edges by using Equation 3 below.
&Quot; (3) "

here,

Stands for structural tensor,

Is the input image, x and y are variables,

Is the derivative of x,

Is the derivative of y.
A computer-readable storage medium having a label area partitioning program stored thereon which functions as a means of any one of claims 1 to 9 so that a label area partitioning method is performed.

Images