Image processing apparatus and image processing method

• US 9,665,803 B2
• Filed: 01/10/2014
• Issued: 05/30/2017
• Est. Priority Date: 01/17/2013
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Aoba, Masato, Saruta, Takayuki
• Primary Examiner(s): Kholdebarin, Iman K
• Application Number:
  US14/152,121
• Publication Number:
  US 20140198951A1
• Time in Patent Office: 1,236 Days
• Field of Search: None
• CPC Class Codes: G06V 20/10 Terrestrial scenes scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59
• CPC Class Codes: G06V 30/242 Division of the character sequences into groups prior to recognition; Selection of dictionaries

Abstract

A dictionary for detection of an object is created from an image obtained by performing an image process, which depends on a first image process parameter, on the training image of the detection target object. The dictionary created based on the image process depending on the first image process parameter is determined, based on a result of detecting the object from an image obtained by performing an image process, which depends on the first image process parameter, on a photographed image based on the dictionary. A second image process parameter is determined, based on a result of detecting the object from an image obtained by performing an image process, which depends on the second image process parameter, on the photographed image using the determined dictionary.

Claims

1. An image processing apparatus comprising:

one or more non-transitory computer-readable storage devices; and

one or more computer processing devices connected to the one or more non-transitory computer-readable storage devices and configured by one or more programs stored in the one or more non-transitory computer-readable storage devices at least to;

set a plurality of first image process parameter candidates;

generate a plurality of processed training images by processing a training image using one of the plurality of first image process parameter candidates, wherein each of the plurality of processed training images is an image generated by using a different first image process parameter candidate;

generate a plurality of dictionary candidates based on the plurality of processed training images;

generate a plurality of first processed test images by processing a first test image using one of the plurality of first image process parameter candidates, wherein each of the plurality of first processed test images is an image generated by using a different first image process parameter candidate;

detect, in each of the plurality of first processed test images, a target object by using one of the plurality of dictionary candidates;

determine a dictionary based on detection results of detecting the target object in the plurality of first processed test images;

set a plurality of second image process parameter candidates;

generate a plurality of second processed test images by processing a second test image using one of the plurality of second image process parameter candidates, wherein each of the plurality of second processed test images is an image generated by using a different second image process parameter candidate;

detect, in each of the plurality of second processed test images, the target object by using the determined dictionary; and

determine a second image process parameter based on detection results of detecting the target object in the plurality of second processed test images; and

store the determined dictionary and the determined second image process parameter in at least one of the one or more non-transitory computer-readable storage devices,

wherein the training image is different from the first and second test images, and

wherein each of the first and second test image is a photographed image.

2. The apparatus according to claim 1, wherein the one or more computer processing devices are further configured by the one or more programs at least to:

set model information of the target object;

calculate a first estimation value associated with a detection error of detecting, in each of the plurality of first processed test images, the target object by using one of the plurality of dictionary candidates, based at least on the model information; and

calculate a second estimation value associated with a detection error of detecting, in each of the plurality of second processed test images, the target object by using the determined dictionary, based at least on the model information.

3. The apparatus according to claim 2, wherein the dictionary is determined based at least on the first estimation value.

4. The apparatus according to claim 2, wherein the second image process parameter is determined based at least on the second estimation value.

5. The apparatus according to claim 2, wherein the training image is a computer graphics image generated from the model information and a luminance distribution of a surface of the target object.

6. The apparatus according to claim 2, wherein the photographed image is an image of the target object captured by an image capturing apparatus.

7. The apparatus according to claim 2,

wherein the calculation of the first estimation value further includes generating a first computer graphics image from the model information based at least on a detecting of the target object in each of the plurality of first processed test images and a comparison of corresponding portions of an edge extracted from the first computer graphics image and an edge extracted from each of the plurality of first processed test images, and

wherein the calculation of the second estimation value further includes generating a second computer graphics image from the model information based at least on a detecting of the target object in each of the plurality of second processed test images and a comparison of corresponding portions of an edge extracted from the second computer graphics image and an edge extracted from each of the plurality of second processed test images.

8. The apparatus according to claim 2,

wherein the calculation of the first estimation value further includes determining first distance information of a predetermined position of the target object from the model information and second distance information of the detected target object in each of the plurality of first processed test images, and comparing the first distance information and the second distance information, and

wherein the calculation of the second estimation value further includes determining third distance information of the detected target object in each of the plurality of second processed test images, and comparing the first distance information and the third distance information.

9. The apparatus according to claim 5, wherein the one or more computer processing devices are further configured by the one or more programs at least to set one or more generation parameters of the training image, and

wherein the computer graphics image is generated according to the one or more generation parameters.

10. The apparatus according to claim 9, wherein each generation parameter of the one or more generation parameters represents a luminance distribution of a surface of the target object.

11. The apparatus according to claim 1, wherein the training image and the photographed image are images captured under image capturing conditions different from each other.

15. The apparatus according to claim 1, wherein each of a first process performed by using one of the first image process parameter candidates and a second process performed by using the second image process parameter or one of the second image process parameter candidates includes at least one of a noise removal process, an edge detection process or a feature detection process.

17. The apparatus according to claim 1, wherein the plurality of first image process parameter candidates and the plurality of second image process parameter candidates are parameters for a same type of image processing.

18. The apparatus according to claim 1, wherein the plurality of first processed test images and the plurality of second processed test images are generated by processing a same test image.

12. An apparatus for recognizing an object from an input image, comprising:

one or more first non-transitory computer-readable storage devices; and

one or more first computer processing devices connected to the one or more non-transitory computer-readable storage devices and configured by one or more programs stored in the one or more non-transitory computer-readable storage devices at least to;

receive an input image;

recognize a target object from the input image according to a dictionary and a second image process parameter determined by an image processing apparatus; and

store a result of the recognition of the target object in at least one of the one or more first non-transitory computer-readable storage devices,

wherein the image processing apparatus comprises;

one or more second non-transitory computer-readable storage devices; and

one or more second computer processing devices connected to the one or more non-transitory computer-readable storage devices and configured by one or more programs stored in the one or more non-transitory computer-readable storage devices at least to;

set a plurality of first image process parameter candidates;

generate a plurality of processed training images by processing a training image using one of the plurality of first image process parameter candidates, wherein each of the plurality of processed training images is an image generated by using a different first image process parameter candidate;

generate a plurality of dictionary candidates based on the plurality of processed training images;

generate a plurality of first processed test images by processing a first test image using one of the plurality of first image process parameter candidates, wherein each of the plurality of first processed test images is an image generated by using a different first image process parameter candidate;

detect, in each of the plurality of first processed test images, a target object by using one of the plurality of dictionary candidates;

determine the dictionary based on detection results of detecting the target object in the plurality of first processed test images;

set a plurality of second image process parameter candidates;

generate a plurality of second processed test images by processing a second test image using one of the plurality of second image process parameter candidates, wherein each of the plurality of second processed test images is an image generated by using a different second image process parameter candidate;

detect, in each of the plurality of second processed test images, the target object by using the determined dictionary;

determine the second image process parameter based on detection results of detecting the target object in the plurality of second processed test images; and

store the determined dictionary and the determined second image process parameter in at least one of the one or more non-transitory computer-readable storage devices,

wherein the training image is different from the first and second test images, and

wherein each of the first and second test image is a photographed image.

13. An image processing method of setting image process parameters defining, at least in part, an image recognition process of detecting a target object from an input image, the method comprising:

causing at least a processor to perform at least the steps of;

setting a plurality of first image process parameter candidates;

generating a plurality of processed training images by processing a training image using one of the plurality of first image process parameter candidates, wherein each of the plurality of processed training images is an image generated by using a different first image process parameter candidate;

generating a plurality of dictionary candidates based on the plurality of processed training images;

generating a plurality of first processed test images by processing a first test image using one of the plurality of first image process parameter candidates, wherein each of the plurality of first processed test images is an image generated by using a different first image process parameter candidate;

detecting, in each of the plurality of first processed test images, a target object by using one of the plurality of dictionary candidates;

determining a dictionary based on detection results of detecting the target object in the plurality of first processed test images;

setting a plurality of second image process parameter candidates;

generating a plurality of second processed test images by processing a second test image using one of the plurality of second image process parameter candidates, wherein each of the plurality of second processed test images is an image generated by using a different second image process parameter candidate;

detecting, in each of the plurality of second processed test images, the target object by using the determined dictionary;

determining a second image process parameter based on detection results of detecting the target object in the plurality of second processed test images; and

storing the determined dictionary and the determined second image process parameter in at least one of the one or more non-transitory computer-readable storage devices,

wherein the training image is different from the first and second test images, and

wherein each of the first and second test image is a photographed image.

14. A non-transitory computer readable storage medium storing a computer-executable program for causing a computer to perform an image processing method of setting image process parameters defining, at least in part, an image recognition process of detecting a target object from an input image, the method comprising:

setting a plurality of first image process parameter candidates;

generating a plurality of processed training images by processing a training image using one of the plurality of first image process parameter candidates, wherein each of the plurality of processed training images is an image generated by using a different first image process parameter candidate;

generating a plurality of dictionary candidates based on the plurality of processed training images;

detecting, in each of the plurality of first processed test images, a target object by using one of the plurality of dictionary candidates;

determining a dictionary based on detection results of detecting the target object in the plurality of first processed test images;

setting a plurality of second image process parameter candidates;

generating a plurality of second processed test images by processing a second test image using one of the plurality of second image process parameter candidates, wherein each of the plurality of second processed test images is an image generated by using a different second image process parameter candidate;

detecting, in each of the plurality of second processed test images, the target object by using the determined dictionary;

determining a second image process parameter based on detection results of detecting the target object in the plurality of second processed test images; and

storing the determined dictionary and the determined second image process parameter in at least one of the one or more non-transitory computer-readable storage devices,

wherein the training image is different from the first and second test images, and

wherein each of the first and second test image is a photographed image.

16. An image processing apparatus comprising:

one or more non-transitory computer-readable storage devices; and

one or more computer processing devices connected to the one or more non-transitory computer-readable storage devices and configured by one or more programs stored in the one or more non-transitory computer-readable storage devices at least to;

read a plurality of dictionary candidates from the one or more non-transitory computer-readable storage devices;

generate a plurality of first processed test images by processing a first test image using one of a plurality of first image process parameter candidates, wherein each of the plurality of first processed test images is an image generated by using a different first image process parameter candidate;

detect, in each of the plurality of first processed test images, a target object by using one of the plurality of dictionary candidates;

determine a dictionary based on detection results of detecting the target object in the plurality of first processed test images;

set a plurality of second image process parameter candidates;

generate a plurality of second processed test images by processing a second test image using one of the plurality of second image process parameter candidates, wherein each of the plurality of second processed test images is an image generated by using a different second image process parameter candidate;

detect, in each of the plurality of second processed test images, the target object by using the determined dictionary;

determine a second image process parameter based on detection results of detecting the target object in the plurality of second processed test images; and

store the determined dictionary and the determined second image process parameter in at least one of the one or more non-transitory computer-readable storage devices,

wherein each of the first and second test image is a photographed image.