Region detecting method and region detecting apparatus

• US 7,657,090 B2
• Filed: 05/26/2004
• Issued: 02/02/2010
• Est. Priority Date: 05/26/2003
• Status: Expired due to Fees

• Current Assignee: Noritsu Koki Co., Ltd.
• Original Assignee: Noritsu Koki Co., Ltd.
• Inventors: Omatsu, Shigeru, Kugo, Kouichi, Mitsukura, Yasue, Kita, Koji, Fukumi, Minoru, Mitsukura, Kensuke
• Primary Examiner(s): Mehta; Bhavesh M
• Assistant Examiner(s): ZARKA, DAVID PETER
• Application Number:
  US10/853,706
• Publication Number:
  US 20040264741A1
• Time in Patent Office: 2,078 Days
• Field of Search: 382/173
• US Class Current: 382/173
• CPC Class Codes: G06T 2207/20012 Locally adaptive
• CPC Class Codes: G06T 2207/30201 Face
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06V 40/162 using pixel segmentation or colour matching

Abstract

To provide a region detecting method capable of setting a proper threshold independently of a photographing condition and moreover, quickly and accurately detecting a specific region, which uses an image data storing section 10 for storing a sample image 12 and a purposed image 11, a first computing means 1 for obtaining the extraction-region-identifying threshold data for a plurality of sample images including a common extraction region but having average color values different from each other in accordance with a genetic algorithm and generating a threshold table 8 for the average color values, a second computing means 2 for adaptively computing the extraction-region-identifying threshold data for the purposed image 11 in accordance with the average color value of the purposed image and the threshold table 8, and a third computing means 3 for detecting an extraction region in accordance with the threshold data computed by the second computing means 12.

Claims

1. A region detecting method based on a method of deciding a threshold value, comprising:

an input step of an input interface section inputting a plurality of sample images and a targeted image and storing the plurality of sample images and the targeted image in a storing section;

a first step of a processor obtaining extraction-region-identifying threshold data for identifying an extraction region for the plurality of sample images including a common extraction region but having average color values different from each other in accordance with an algorithm;

said first step including

a threshold data group generating step of the processor initially generating a predetermined number of threshold data groups constituted by at least RGB threshold data for a first color-component region and RGB threshold data for a second color-component region adjacent to the first color-component region for characterizing the extraction region as the extraction-region-identifying threshold data for identifying the extraction region,

an adaptability evaluating step of the processor evaluating an adaptability for the sample images of the generated threshold data groups,

a gene processing step of the processor generating new threshold data groups by executing at least one of a crossing operation and a mutation operation and selecting the new threshold data groups in accordance with the adaptability evaluated in said adaptability evaluating step,

a threshold data group supplying step of the processor generating additional threshold data groups to maintain a number of the threshold data groups equal to a number of the threshold data groups generated in said threshold data group generating step, and

a generation alternating step of the processor repeating said gene processing step, said adaptability evaluating step and said threshold data group supplying step over set generations;

a second step of the processor adaptively computing the extraction-region-identifying threshold data for identifying the extraction region for the targeted image in accordance with average color values of the targeted image and a relation between the thresholds and the average color values obtained in said first step;

a third step of the processor detecting the extraction region in accordance with the threshold data computed in said second step; and

an output step of the processor outputting the extraction region detected in said third step to an output interface section.

2. A region detecting apparatus based on a method of deciding a threshold value, comprising:

an input interface section for inputting a plurality of sample images and a targeted image and storing the plurality of sample images and the targeted image in the storing section;

a first computing unit obtaining extraction-region-identifying threshold data for identifying an extraction region for the plurality of sample images including a common extraction region but having average color values different from each other in accordance with an algorithm and generating a threshold table for the average color values;

said first computing unit including

a threshold data group generating unit initially generating threshold data groups constituted by at least RGB threshold data for a first color-component region and RGB threshold data for a second color-component region adjacent to the first color-component region for characterizing the extraction region,

an adaptability evaluating unit evaluating an adaptability for the sample images of the generated threshold data groups,

a gene processing unit generating new threshold data groups by executing at least one of a crossing operation and a mutation operation and selecting the new threshold data groups in accordance with the adaptability evaluated in said adaptability evaluating unit,

threshold data group supplying unit generating additional threshold data groups to maintain a number of the threshold data groups equal to a number of the threshold data groups generated in said threshold data group generating unit, and

a generation alternation computing unit repeating operations performed by said gene processing unit, said adaptability evaluating unit and said threshold data group supplying unit over set generations;

a second computing unit adaptively computing the extraction-region-identifying threshold data for identifying the extraction region for the targeted image in accordance with average color value of the targeted image and a threshold table;

a third computing unit detecting the extraction region in accordance with the threshold data computed by said second computing unit; and

an output interface section for outputting the extraction region detected by said third computing unit to a computer.

3. A region detecting apparatus based on a method of deciding a threshold value, comprising:

an input interface section for inputting a plurality of sample images and a targeted image and storing the plurality of sample images and the targeted image in the storing section;

a first computing unit obtaining extraction-region-identifying threshold data for identifying an extraction region for the plurality of sample images including a common extraction region but having average color values different from each other in accordance with an algorithm and generating a threshold table for the average color values;

said first computing unit including

a threshold data group generating unit initially generating a predetermined number of threshold data groups constituted by at least RGB threshold data for a first color-component region and RGB threshold data for a second color-component region adjacent to the first color-component region for characterizing the extraction region,

an adaptability evaluating unit evaluating an adaptability for the sample images of the generated threshold data groups,

a gene processing unit generating new threshold data groups by executing at least one of a crossing operation and a mutation operation and selecting the new threshold data groups in accordance with the adaptability evaluated in said adaptability evaluating unit,

a threshold data group supplying unit generating additional threshold data groups to maintain a number of the threshold data groups equal to a number of the threshold data groups generated in said threshold data group generating unit, and

a generation alternation computing unit repeating operations performed by said gene processing unit, said adaptability evaluating unit and said threshold data group supplying unit over set generations;

a second computing unit adaptively computing the extraction-region-identifying threshold data for identifying the extraction region for the targeted image in accordance with average color value of the targeted image and a threshold table;

a third computing unit detecting the extraction region in accordance with the threshold data computed by said second computing unit; and

an output interface section for outputting the extraction region detected by said third computing unit.