Extracting feature quantities from an image to perform location estimation

• US 9,020,196 B2
• Filed: 09/10/2013
• Issued: 04/28/2015
• Est. Priority Date: 08/15/2012
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Takahashi, Toshihiro
• Primary Examiner(s): Koziol, Stephen R
• Assistant Examiner(s): SAINI, AMANDEEP SINGH
• Application Number:
  US14/022,368
• Publication Number:
  US 20140050359A1
• Time in Patent Office: 595 Days
• Field of Search: None
• US Class Current: 382/103
• CPC Class Codes: G06T 7/579 from motion
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06V 30/142 using hand-held instruments; Constructional details of the instruments

Abstract

A feature extraction method for extracting a feature from an image includes receiving an image and measured acceleration data from a mobile device; obtaining a gravity vector in the image in a camera coordinate system based on the measured acceleration data; obtaining a vanishing point in the image in a vertical direction in a screen coordinate system using the gravity vector; obtaining differential vectors along two axes for each pixel in the screen coordinate system; obtaining a connection line vector connecting each of the pixels with the vanishing point; identifying a vertical edge based on determining that an angle formed by the differential vector and the connection line vector is within a certain threshold range; obtaining the sum of strengths of vertical edges and writing the sum in a predetermined variable array; extracting a keypoint based on the variable array; and calculating a feature quantity from the keypoint.

Claims

1. A feature extraction system for extracting a feature from an image by computer processing, the system comprising:

means for receiving an image and measured acceleration data from a mobile device including image acquiring means and acceleration measuring means;

means for obtaining a gravity vector in the image in a camera coordinate system on the basis of the measured acceleration data;

means for obtaining a vanishing point in the image in a vertical direction in a screen coordinate system by using the gravity vector;

means for obtaining differential vectors along two axes for each of pixels in the screen coordinate system;

means for obtaining a connection line vector connecting each of the pixels with the vanishing point;

means for identifying a vertical edge on the basis of determination that an angle formed by the differential vector and the connection line vector is within a certain threshold range;

means for obtaining the sum of strengths of vertical edges and writing the sum in a predetermined variable array;

means for extracting a keypoint on the basis of the variable array; and

means for calculating a feature quantity from the keypoint.

2. The system according to claim 1, wherein the mobile device is a smartphone.

3. The system according to claim 2, wherein the acceleration measuring means is an acceleration sensor.

4. The system according to claim 1, wherein the means for obtaining the sum of strengths of vertical edges and writing the sum in a predetermined variable array performs processing for calculating the absolute value of the inner product of the differential vector and (u′

−

u, v′

−

v) and writing the absolute value in the variable array, where (u, v) is screen coordinates and (u′

, v′

) is the vanishing point.

5. The system according to claim 1, wherein the means for extracting a keypoint performs processing for assuming the variable array to be a horizontally long image and calculating an extremum by using a DoG (difference of Gaussian) function.

6. The system according to claim 1, wherein the means for calculating the feature quantity performs processing for calculating HoG (Histograms of Oriented Gradients).