IMAGE PROCESSING DEVICE AND IMAGE PROCESSING METHOD

• US 20140152780A1
• Filed: 10/01/2013
• Published: 06/05/2014
• Est. Priority Date: 11/30/2012
• Status: Abandoned Application

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: NAKAYAMA, Osafumi, KANETAKE, Jun
• Application Number:
  US14/043,439
• Publication Number:
  US 20140152780A1
• Time in Patent Office: Days
• US Class Current: 348/50
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20012 Locally adaptive
• CPC Class Codes: G06T 2207/30264 Parking
• CPC Class Codes: G06T 7/579 from motion
• CPC Class Codes: H04N 13/282 for generating image signals corresponding to three or more geometrical viewpoints, e.g. multi-view systems

Abstract

An image processing apparatus includes an acquiring unit for acquiring an image from a camera in a moving body; an extracting unit for extracting feature points from the image; a matching unit for performing a matching process on feature points extracted from images taken at different time points; a position calculating unit for calculating a three-dimensional position based on the matching feature point and the movement of the moving body; a calculating unit for calculating a precision of the three-dimensional position; a distribution determining unit for detecting an object from the image and setting a threshold for each object based on a precision distribution of the feature points of each object; a selecting unit for selecting, for each object, the feature points having a higher precision than the threshold; and a generating unit for generating an object shape by using the feature points that have been selected.

Claims

1. An image processing apparatus comprising:

an image acquiring unit configured to acquire an image from a camera installed in a moving body;

a feature point extracting unit configured to extract feature points from the image;

a matching unit configured to perform a matching process on the feature points extracted from the images that have been taken at different time points;

a position calculating unit configured to calculate a three-dimensional position based on a position of the feature point that has been matched and a movement amount of the moving body;

a precision calculating unit configured to calculate a precision indicating an error of the three-dimensional position;

a distribution determining unit configured to detect an object from the image and set a threshold for each object based on a precision distribution of the feature points generated for each object;

a selecting unit configured to select, for each object, the feature points having a higher precision than the threshold; and

a generating unit configured to generate an object shape by using the feature points that have been selected.

2. The image processing apparatus according to claim 1, further comprising:

a precision predicting unit configured to predict the precision of the three-dimensional position based on a movement of the moving body, wherein

the selecting unit determines the feature points to be selected based on a prediction result of the precision predicting unit.

3. The image processing apparatus according to claim 2, wherein

the selecting unit selects the feature points by using a common threshold, and

selects the feature points by using the threshold set for each object, from among the feature points that have not been selected and whose precision is predicted to decrease from a present precision by the precision predicting unit.

4. The image processing apparatus according to claim 1, further comprising:

a distance calculating unit configured to obtain a feature point that is closest to the moving body among the feature points of each object, wherein

the selecting unit selects the feature points from among the feature points that are within a predetermined range from the feature point closest to the moving body.

5. The image processing apparatus according to claim 1, further comprising:

a distance determining unit configured to determine whether a distance from the moving body to the feature point of the object is greater than or equal to a threshold distance, based on the three-dimensional position, wherein

a process of a latter stage performed from and beyond the precision calculating unit is performed on the feature points determined to be at less than or equal to the threshold distance by the distance determining unit.

6. An image processing method executed by a computer, the image processing method comprising:

acquiring an image from a camera installed in a moving body;

extracting feature points from the image;

performing a matching process on the feature points extracted from the images that have been taken at different time points;

calculating a three-dimensional position based on a position of the feature point that has been matched and a movement amount of the moving body;

calculating a precision indicating an error of the three-dimensional position;

detecting an object from the image and setting a threshold for each object based on a precision distribution of the feature points generated for each object;

selecting, for each object, the feature points having a higher precision than the threshold; and

generating an object shape by using the feature points that have been selected.

7. A non-transitory computer-readable recording medium storing a program that causes a computer to execute a process comprising:

acquiring an image from a camera installed in a moving body;

extracting feature points from the image;

performing a matching process on the feature points extracted from the images that have been taken at different time points;

calculating a three-dimensional position based on a position of the feature point that has been matched and a movement amount of the moving body;

calculating a precision indicating an error of the three-dimensional position;

detecting an object from the image and setting a threshold for each object based on a precision distribution of the feature points generated for each object;

selecting, for each object, the feature points having a higher precision than the threshold; and

generating an object shape by using the feature points that have been selected.