Image search apparatus, image search method, and storage medium for matching images with search conditions using image feature amounts

• US 8,532,431 B2
• Filed: 04/24/2008
• Issued: 09/10/2013
• Est. Priority Date: 05/08/2007
• Status: Active Grant

• Current Assignee: Canon Ayutthaya Limited (Canon Inc.)
• Original Assignee: Canon Ayutthaya Limited (Canon Inc.)
• Inventors: Hori, Shinjiro
• Primary Examiner(s): Vital, Pierre
• Assistant Examiner(s): Raab, Christopher J
• Application Number:
  US12/109,255
• Publication Number:
  US 20080281797A1
• Time in Patent Office: 1,965 Days
• Field of Search: 382/305, 382/151, 382/218, 382/289, 382/293, 382/294, 382/295, 382/298, 382/277, 707/705, 707/736
• US Class Current: 382/277
• CPC Class Codes: G06F 16/5854 using shape and object relationship

Abstract

With the conventional techniques, when the images taken under various conditions such as plural aspect ratios, horizontal shooting and vertical shooting are collectively selected and searched, the search conditions adaptable to all the images can not be specified, whereby there was a problem that the search could not be made as intended. In order to solve this problem, the present invention provides an image search apparatus comprising search object image designation means for designating a search object image, search condition input means for inputting the search conditions, feature amount extraction means for extracting a feature amount thereof from the search object image, feature amount transformation means for transforming the feature amount to adapt to the inputted search conditions, and determination means for determining whether or not the search object image is matched with the search conditions using the feature amount after transformation.

Claims

1. An image search apparatus comprising:

search object image designation means for designating search object images;

search condition input means for inputting search conditions;

for each designated search object image;

feature amount extraction means for extracting a feature amount from the search object image;

feature amount transformation means for transforming said extracted feature amount to adapt to said input search conditions; and

determination means for determining whether or not said search object image is matched with the search conditions using said transformed feature amount after transformation,

wherein said feature amount transformation means comprises coordinate transformation means for transforming coordinates of a noticed subject included in said extracted feature amount into coordinates in a coordinate system as defined by said search conditions,

wherein said coordinate transformation means transforms a long-side coordinate interval of a long-side direction of the coordinate system for the search object image,

wherein for a long-side coordinate interval that is within a predetermined range from a center of the coordinate system for the search object image, said coordinate transformation means transforms the long-side coordinate interval to be similar to a coordinate interval in a short-side direction,

wherein for a long-side coordinate interval that is outside the predetermined range, said coordinate transformation means transforms the long-side coordinate interval to be wider than the coordinate interval in said short-side direction,

wherein said determination means makes said determination by making greater a tolerance of determination near the periphery than at the center of the coordinate system,

wherein a long-side coordinate interval that is near the periphery of the coordinate system is outside the predetermined range, and a long-side coordinate interval that is at the center of the coordinate system is within the predetermined range,

wherein the long-side coordinate interval near the periphery of the coordinate system is transformed differently from the long-side coordinate interval at the center of the coordinate system, and

wherein said determination means makes greater the tolerance of match determination near the periphery of the coordinate system in consideration of input errors of the search conditions near the periphery that are generated by transforming the long-side coordinate interval near the periphery to be wider than the coordinate interval in said short-side direction.

2. The image search apparatus as claimed in claim 1, wherein said coordinate transformation means makes the center of the coordinate system for said search object image coincide with the center of the coordinate system as defined by said search conditions in making said transformation.

3. The image search apparatus as claimed in claim 1, wherein said coordinate transformation means makes the end of the coordinate system for said search object image coincide with the end of the coordinate system as defined by said search conditions in making said transformation.

4. The image search apparatus as claimed in claim 1, wherein said search condition input means accepts the inputs of designating the position, size and inclination of the noticed subject.

5. An image search method for use in an image search apparatus, the method comprising:

a search object image designation step of designating search object images;

a search condition input step of inputting search conditions;

for each designated search object image;

a feature amount extraction step of extracting a feature amount from the search object image;

a feature amount transformation step of transforming said extracted feature amount to adapt to said input search conditions; and

a determination step of determining whether or not said search object image is matched with the search conditions using said transformed feature amount after transformation,

wherein said feature amount transformation step includes a coordinate transformation step of transforming coordinates of a noticed subject included in said extracted feature amount into coordinates in a coordinate system as defined by said search conditions,

wherein said coordinate transformation step transforms a long-side coordinate interval of a long-side direction of the coordinate system for the search object image,

wherein for a long-side coordinate interval that is within a predetermined range from a center of the coordinate system for the search object image, said coordinate transformation step transforms the long-side coordinate interval to be similar to a coordinate interval in a short-side direction,

wherein for a long-side coordinate interval that is outside the predetermined range, said coordinate transformation step transforms the long-side coordinate interval to be wider than the coordinate interval in said short-side direction,

wherein said determination step makes said determination by making greater a tolerance of determination near the periphery than at the center of the coordinate system,

wherein a long-side coordinate interval that is near the periphery of the coordinate system is outside the predetermined range, and a long-side coordinate interval that is at the center of the coordinate system is within the predetermined range,

wherein the long-side coordinate interval near the periphery of the coordinate system is transformed differently from the long-side coordinate interval at the center of the coordinate system, and

wherein said determination step makes greater the tolerance of match determination near the periphery of the coordinate system in consideration of input errors of the search conditions near the periphery that are generated by transforming the long-side coordinate interval near the periphery to be wider than the coordinate interval in said short-side direction.

6. The image search method as claimed in claim 5, wherein said coordinate transformation step includes making the center of the coordinate system for said search object image coincide with the center of the coordinate system as defined by said search conditions in making said transformation.

7. The image search method as claimed in claim 5, wherein said coordinate transformation step includes making the end of the coordinate system for said search object image coincide with the end of the coordinate system as defined by said search conditions in making said transformation.

8. The image search method as claimed in claim 5, wherein said search condition input step includes accepting the inputs of designating the position, size and inclination of the noticed subject.

9. A non-transitory computer readable storage medium that retrievably stores a program for causing a computer to perform an image search method, the method comprising:

a search object image designation step of designating search object images;

a search condition input step of inputting search conditions;

for each designated search object image;

a feature amount extraction step of extracting a feature amount from the search object image;

a feature amount transformation step of transforming said extracted feature amount to adapt to said input search conditions; and

a determination step of determining whether or not said search object image is matched with the search conditions using said transformed feature amount after transformation,

wherein said feature amount transformation step includes a coordinate transformation step of transforming coordinates of a noticed subject included in said extracted feature amount into coordinates in a coordinate system as defined by said search conditions,

wherein said coordinate transformation step transforms a long-side coordinate interval of a long-side direction of the coordinate system for the search object image,

wherein for a long-side coordinate interval that is within a predetermined range from a center of the coordinate system for the search object image, said coordinate transformation step transforms the long-side coordinate interval to be similar to a coordinate interval in a short-side direction,

wherein for a long-side coordinate interval that is outside the predetermined range, said coordinate transformation step transforms the long-side coordinate interval to be wider than the coordinate interval in said short-side direction,

wherein said determination step makes said determination by making greater a tolerance of determination near the periphery than at the center of the coordinate system,

wherein a long-side coordinate interval that is near the periphery of the coordinate system is outside the predetermined range, and a long-side coordinate interval that is at the center of the coordinate system is within the predetermined range,

wherein the long-side coordinate interval near the periphery of the coordinate system is transformed differently from the long-side coordinate interval at the center of the coordinate system, and

wherein said determination step makes greater the tolerance of match determination near the periphery of the coordinate system in consideration of input errors of the search conditions near the periphery that are generated by transforming the long-side coordinate interval near the periphery to be wider than the coordinate interval in said short-side direction.