Line segment detection method

• US 10,748,026 B2
• Filed: 12/19/2017
• Issued: 08/18/2020
• Est. Priority Date: 10/09/2015
• Status: Active Grant

• Current Assignee: IHI Corporation
• Original Assignee: IHI Corporation
• Inventors: Sonoda, Shinichirou, Banno, Hajime, Mizouchi, Kentaro
• Primary Examiner(s): Woldemariam, Aklilu K
• Application Number:
  US15/846,502
• Publication Number:
  US 20180121749A1
• Time in Patent Office: 973 Days
• Field of Search: 348142, 348145, 348148, 382103, 382113, 382173, 382276
• CPC Class Codes: G06T 2207/10032 Satellite or aerial image; Remote sensing
• CPC Class Codes: G06T 2207/20061 Hough transform
• CPC Class Codes: G06T 2207/30241 Trajectory
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation
• CPC Class Codes: G06T 7/70 Determining position or orientation of objects or cameras camera calibration G06T7/80
• CPC Class Codes: G06T 7/75 involving models

Abstract

A line segment detection method includes acquiring an image, extracting from the image a straight line that includes a line segment, acquiring a luminance distribution of the straight line, generating a virtual function model that contains a distribution of a foreground and a background of the image by using the luminance distribution, and determining whether luminance values near end points (start point and end point) of the line segment contained in the virtual function model probabilistically belong to the foreground or the background, and estimating end points (start point and end point) in the virtual function model.

Claims

1. A line segment detection method for detecting a line segment from an image, the line segment detection method comprising:

extracting a straight line including the line segment from the image;

acquiring a luminance distribution on the straight line;

generating a virtual function model that contains a distribution of a foreground and a background of the image based on the luminance distribution, wherein the foreground is part of the straight line assumed as the line segment within a start point and an end point virtually set at arbitrary positions on the straight line, and the background is part of the straight line except for the foreground on the straight line;

estimating values of the end points of the line segment in the virtual function model by determining whether luminance values near the start point and the end point on the straight line probabilistically belong to the foreground or the background,

repeatedly regenerating the virtual function model by using the estimated values of the end points obtained at the estimating the end points,

wherein the virtual function model is any one of a likelihood function model, a ramp function model, and a sigmoid function model.

2. The line segment detection method according to claim 1 further comprising judging whether a detection is erroneous or not by using the estimated values of the end points obtained at the estimating the end points.

3. The line segment detection method according to claim 1, wherein the estimating the end points includes estimating the end points by using a maximum likelihood method.

4. The line segment detection method according to claim 1, wherein the image contains debris that moves in the Earth'"'"'s orbit, and the line segment is a movement trajectory of the debris.