US 12,731,236 B2Granted

Method of measuring structure displacement based on fusion of asynchronous vision measurement data of natural target and acceleration data of structure, and system for the same

AssigneeKorea Advanced Institute of Science and Technology

InventorsHoon Sohn, Jae Mook Choi, Zhanxiong Ma, Ki Young Kim

Life of this patent

Estimating the term…

Loading patent…

Abstract

Disclosed are a method and a system for measuring a displacement of a structure based on fusion of asynchronous vision measurement data of natural target and acceleration data of the structure. A scale factor, which is a conversion factor between a pixel resolution of the image frame data and a distance resolution of real space, is calculated using image frame data taken at a predetermined first sample frequency of a stationary target outside the structure by a camera installed on the structure, and acceleration data measured at a predetermined second sample frequency by an accelerometer installed on the structure, over time. A vision-based displacement value of the structure is estimated using a feature point matching algorithm and the scale factor.

Claims16

Dependent claims

A method of measuring a displacement of a structure, comprising:

calculating a scale factor (α), which is a conversion factor between a pixel resolution of the image frame data and a distance resolution of real space, by a processor, using a series of image frame data taken at a predetermined first sample frequency of a stationary target outside the structure by a camera installed on the structure to be measured over time, and a series of acceleration data measured at a predetermined second sample frequency by an accelerometer installed on the structure;
and
calculating a vision-based displacement value of the structure, by the processor, using a feature point matching algorithm and the scale factor, and
computing a displacement estimate of the structure at a sample frequency of the acceleration data, by the processor, by fusing asynchronous the image frame data and the acceleration data to be synchronized with each other with noise reduced using an adaptive multi-rate Kalman filter while using the vision-based displacement value,
wherein the ‘computing a displacement estimate of the structure’ comprises updating a region of interest (ROI) of an ith image frame;
matching N feature points within a ROI of a first image frame with N feature points within the ROI of the ith image frame using a predetermined feature point matching algorithm;
rejecting feature point mismatches to leave only good feature point matches among the N feature point matches using a mismatch rejection algorithm;
and computing a final translation estimate by using an equation of
u
t
=
α
⁢
{
1
N
g
⁢
(
∑
j
=
1
N
⁢
g
⁢
d
i
j
)
+
d
i
ROI
}
,
where Ng denotes number of the good feature point matches, in which the scale factor (α) is applied to a sum of the translations of the good feature point matches and movements of the ROIs.

35 more elements

A system for measuring a displacement of a structure, comprising:

a camera installed on the structure, configured to image a stationary target outside the structure at a predetermined first sample frequency;
an accelerometer collocated with the camera on the structure, configured to measure acceleration at a predetermined second sample frequency;
and
a displacement estimation unit, implemented in a processor, comprising a scale factor calculation unit, and an adaptive multi-rate Kalman filter unit, wherein the scale factor calculation unit is configured to perform a function of calculating a scale factor (α), which is a conversion factor between a pixel resolution of the image frame data and a distance resolution of real space, using a series of image frame data taken at a predetermined first sample frequency of a stationary target outside the structure by the camera and a series of acceleration data measured at a predetermined second sample frequency by the accelerometer over time, and wherein the adaptive multi-rate Kalman filter unit is configured to perform functions of calculating a vision-based displacement value of the structure using a feature point matching algorithm and the scale factor;
and computing a displacement estimate of the structure at a sampling frequency of the acceleration data by fusing asynchronous the acceleration data and the image frame data to be synchronized with each other with noise reduced using an adaptive multi-rate Kalman filter while using the vision-based displacement value,
wherein the function of ‘computing a displacement estimate of the structure’, comprises:
updating a region of interest (ROI) of an ith image frame;
matching N feature points within a ROI of a first image frame with N feature points within the ROI of the ith image frame using a predetermined feature point matching algorithm;
rejecting feature point mismatches to leave only good feature point matches among the N feature point matches using a mismatch rejection algorithm;
and computing a final translation estimate by using an equation of
u
t
=
α
⁢
{
1
N
g
⁢
(
∑
j
=
1
N
⁢
g
⁢
d
i
j
)
+
d
i
ROI
}
,
where Ng denotes number of the good feature point matches, in which the scale factor (α) is applied to a sum of the translations of the good feature point matches and movements of the ROIs.

37 more elements

Dependencies are inferred from the claim text.