Image recognition device, image recognition method, program, and recording medium

• US 9,747,511 B2
• Filed: 03/27/2013
• Issued: 08/29/2017
• Est. Priority Date: 11/16/2012
• Status: Active Grant

• Current Assignee: Mitsubishi Heavy Industries Machinery Systems, Ltd. (Mitsubishi Heavy Industries, Ltd.)
• Original Assignee: Mitsubishi Heavy Industries, Ltd.
• Inventors: Kojima, Yohei, Sugimoto, Kiichi, Okazaki, Takuma, Nakao, Kenta
• Primary Examiner(s): BHATNAGAR, ANAND P
• Application Number:
  US14/442,715
• Publication Number:
  US 20160132742A1
• Time in Patent Office: 1,616 Days
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06V 20/62 Text, e.g. of license plates, overlay texts or captions on TV images
• CPC Class Codes: G06V 20/625 License plates
• CPC Class Codes: G06V 20/63 Scene text, e.g. street names

Abstract

An image recognition device includes an edge direction evaluation unit configured to evaluate a direction of an edge at each coordinate by assigning points to the coordinate when a deviation amount between a direction of an outline of an abundance ratio higher than a predetermined abundance ratio threshold value among directions of an outline of a character available for the vehicle number plate and the direction of the edge at each coordinate of the image is less than a predetermined deviation amount threshold value and a vehicle number plate region extraction unit configured to extract a region of the vehicle number plate from the image based on a distribution of the points assigned by the edge direction evaluation unit.

Claims

1. An image recognition device for extracting a region of a vehicle number plate from an image, the image recognition device comprising:

an edge direction evaluation unit configured to evaluate a direction of an edge at each coordinate by assigning points to the coordinate when a deviation amount between a direction of an outline of an abundance ratio higher than a predetermined abundance ratio threshold value among directions of an outline of a character available for the vehicle number plate and the direction of the edge at each coordinate of the image is less than a predetermined deviation amount threshold value; and

a vehicle number plate region extraction unit configured to extract the region of the vehicle number plate from the image based on a distribution of the points assigned by the edge direction evaluation unit.

2. The image recognition device according to claim 1, further comprising:

an edge intensity determination unit configured to determine whether an intensity of the edge at each coordinate of the image is greater than a predetermined intensity threshold value,

wherein the edge direction evaluation unit does not assign points to the coordinate even when the deviation amount between the direction of the outline and the direction of the edge is less than the predetermined deviation amount threshold value if the edge intensity determination unit determines that the intensity of the edge is less than the predetermined intensity threshold value.

3. The image recognition device according to claim 1, further comprising:

a horizontal distribution totalization unit configured to totalize a distribution of the horizontal direction of the points assigned by the edge direction evaluation unit,

wherein the vehicle number plate region extraction unit extracts a region of the horizontal direction of the vehicle number plate from the image based on the distribution of the points of the horizontal direction totalized by the horizontal distribution totalization unit.

4. The image recognition device according to claim 1, further comprising:

a vertical distribution totalization unit configured to totalize a distribution of a vertical direction of the points assigned by the edge direction evaluation unit,

wherein the vehicle number plate region extraction unit extracts a region of the vertical direction of the vehicle number plate from the image based on the distribution of the points of the vertical direction totalized by the vertical distribution totalization unit.

5. The image recognition device according to claim 1, further comprising:

a regional distribution totalization unit configured to totalize a distribution of each predetermined region of the points assigned by the edge direction evaluation unit,

wherein the vehicle number plate region extraction unit extracts the region of the vehicle number plate from the image based on the distribution of the points of each predetermined region totalized by the regional distribution totalization unit.

6. The image recognition device according to claim 1, wherein the edge direction evaluation unit assigns points differing according to the abundance ratio of the direction of the outline when the deviation amount between the direction of the outline and the direction of the edge is less than the predetermined deviation amount threshold value.

7. An image recognition method of extracting a region of a vehicle number plate from an image, the image recognition method comprising:

an edge direction evaluation step of evaluating a direction of an edge at each coordinate by assigning points to the coordinate when a deviation amount between a direction of an outline of an abundance ratio higher than a predetermined abundance ratio threshold value among directions of an outline of a character available for the vehicle number plate and the direction of the edge at each coordinate of the image is less than a predetermined deviation amount threshold value; and

a vehicle number plate region extraction step of extracting the region of the vehicle number plate from the image based on a distribution of the points assigned in the edge direction evaluation step.

8. A non-transitory computer readable medium containing a program for causing a computer to function as an image recognition device for extracting a region of a vehicle number plate from an image, the computer being caused to function as:

an edge direction evaluation unit configured to evaluate a direction of an edge at each coordinate by assigning points to the coordinate when a deviation amount between a direction of an outline of an abundance ratio higher than a predetermined abundance ratio threshold value among directions of an outline of a character available for the vehicle number plate and the direction of the edge at each coordinate of the image is less than a predetermined deviation amount threshold value; and

a vehicle number plate region extraction unit configured to extract the region of the vehicle number plate from the image based on a distribution of the points assigned by the edge direction evaluation unit.