Feature Tracking using Close-Range Photogrammetry

Feature Tracking using Close-Range Photogrammetry

Presentation

Under the supervision of Dr. Saurabh Vijay

By- Samraddhi Prajapati Sakshi Singh Shrey Anand Sachin Singh

Author: Name Surname Advisor: Name Surnam Date: 20XX

Index

1.

Introduction

2.

Objective

3.

Timeline

4.

Softwares

5.

Methodology

6.

Result

7.

Conclusions

Introduction

• Feature tracking using close-range photogrammetry is a technique used to accurately measure the movement of objects and surfaces over time.
• This technique involves taking a series of photographs of an object from multiple angles using a camera and then using software to track specific features in each image.
• By tracking these features over time, it is possible to create a 3D model of the surface and measure its movement with high precision.

+ info

Objectives

To calculate real-world measurements based on time-lapse oblique images.

Calculation of velocity on the real word data model using Pytrx and ImGRAFT

Exploring various software on their sample data and documenting necessary parameters for softwares.

Tool box Exploration

Velocity calculation

+ info

+ info

ImGRAFT

PyTrx

It is an open-source MATLAB package that combines image geo-rectification and feature tracking post-processing steps into a comprehensive toolbox.

PyTrx is a Python-based open-source software tool for the automated tracking of moving objects.

Softwares

There are many software packages available that can perform these tasks, each with its own set of features, strengths, and limitations.

PIVlabs

EMT

Particle Image Velocimetry (PIV) is technique to measure fluid velocities.

EMT is a free research software that offers a comprehensive workflow for the analysis of time-lapse imagery.

ImGRAFT

Image georectification and feature tracking toolbox:

• It is a free and open-source image georectification and feature tracking toolbox (ImGRAFT), which we employ in our data to make a velocity field from a monoscopic, oblique image.

• It assimilates the rectification of the images and subsequent feature tracking into one toolbox.

These features are advantageous, as current existing software tends to address only one of these processes, and thereby require numerous software to complete the entire processing.

+ info

Timeline

March

January

May

Real-world data acquisition was done.

Different Toolbox were explored.

Final Report

February

April

Sample data execution was done.

Implementation of different software over same Data.

PyTrx

Python Tracking

PyTrx is a Python-based open-source software tool for the automated tracking of moving objects. It is designed to operate with Python 3. It provides an easy-to-use interface for users. It has been developed with the primary purpose of analyzing glaciological data. It provides users with a range of functions for analyzing glacier surface velocities, lake surface areas, and glacial line profiles using both automated and manual methods. One of PyTrx's main features is to derive surface velocities using two feature-tracking methods.

Workflow of PyTrx

EMT

Environmental Motion Tracking

• EMT is a free research software that enables the analysis of time-lapse imagery. It offers features like motion tracking, camera motion correction, and scaling and geo-referencing capabilities, making it ideal for accurate analysis of image changes over time.

• EMT has limitations in georeferencing functionality, dem calculator error and unavalability of sample data .
• Users should consider these constraints before utilizing EMT for time-lapse imagery analysis.

+ info

PIVlab

Particle image velocimetry (PIV) tool with GUI

PIVlab is a software in Matlab using a graphical user interface. Particle Image Velocimetry (PIV) is a technique to measure fluid velocities. PIV is a method that is used to get space-resolved velocity information or displacement measurement from image data.

PIVlab Cont.

Calibration in PIVlab was different from other software as it does not require GCP and DEM.

PIVlab is primarily designed for 2D PIV analysis and does not have built-in support for 3D PIV analysis.

Sample Data (PIVlabs)

Study Area

Based on accessibility, environmental conditions, and available resources. We choose Hockey Ground for data acquisition. Several factors have influenced our decision:

• During the daytime, there is no interruption from humans.
• There is an open area for flying drone.
• Large area with a width of up to 60 metres.

Data and Equipment Used

Table

You can write a description here

Data Acquistion

Post-Processing

Methodology

Data Acquisition 2

Next Day, Drone flown to obtain Digital Elevation Model (DEM) data

• Gps Base installed at Geomatics building.

• Drone calibrated before flight to ensure accurate flying.

• GPS rover installed on drone for accurate positioning during flight.

• Drone covered 0.007 sq. km and captured 9 images for DEM data

Data Acquisition 1

• Gps Base installed at Geomatics building.
• Set up camera to capture the desired field of view
• Placed 10 ground control points (GCPs) marked with white bricks for camera calibration.
• Took GPS readings of GCPs to accurately record their locations.
• Placed 5 red bricks in different locations to distinguish moving objects from GCPs.
• Displacement of red bricks measured in 5 iterations using meter tape
• Camera captured images at 5-minute intervals

Post Processing

DEM Preparation

GPS POSTPROCESSING

CAMERA CALIBRATION

Processing done after Data-Acquistion in Lab.

Extrinsic and Intrinsic MATLAB Camera-Calibrator Nine-Images Required (Chess Box)

Two softwares were used for the preparation of DEM, one is Pix4Dmapper and other one is Emild Studio.

DGPS system base and rover reading from the receiverSpectrum survey software used

+ info

+ info

+ info

Methodology

PyTrx :

• The methodology of PyTrx can be divided into several key steps:
• The first step is to acquire multiple images and filter them out like removing images that have any interference in them.
• The next step is to use GCPs data to calculate camera parameters, these parameters include both internal and external parameters.
• In order to convert the pixel measurements into real world values, we use a DEM with other data to convert the parameters into a reference image.

Result

PyTrx :

• In this study, close-range photogrammetry was used to track features such as object motion, and the results were analyzed to determine the accuracy and precision of different software tools.
• While running the acquired data we came across many errors that were basically related to the compatibility of PyTrx. Since our data was highly intensive, we modified our DEM and changed its resolution. The next error was related to the limitation of PyTrx not having any functions for the removal of shadow from images.
• The results show two images, the first image shows the tracked static points in red color and the next image shows the tracked velocity. The results are not accurate due to the shadow removal limitation of PyTrx.

Tracked static points (Red points)

Tracked velocity

Conclusions

Resolution of DEM and accuracy of GPS plays an vital role for final feature Detection.

It is used for qualitative and quantitative analysis of time lapse imagery .

Close-Range Photogrammetry

DEM & GPS

+ info

+ info

For tracking analysis, Pytrx and Imgraft suited best as nominal parameters along with easy interface and result analysis

Having done a project we have gathered information about key parameters that can be applied to larger projects.

Experience

PyTrx & ImGRAFT

+ info

+ info

Thank you!

References

Bibliography

Messerli, A. and Grinsted, A. (2015), Image GeoRectification And Feature Tracking toolbox: ImGRAFT, Geosci. Instrum. Method. Data Syst., 4, 23-34, doi:10.5194/gi-4-23-2015

Schwalbe, Ellen; Maas, Hans-Gerd: The determination of high-resolution spatio-temporal glacier motion fields from time-lapse sequences. In: Journal of Earth Surface Dynamics, 5 (2017), S. 861–879

Schwalbe, Ellen; Maas, Hans-Gerd: The determination of high-resolution spatio-temporal glacier motion fields from time-lapse sequences. In: Journal of Earth Surface Dynamics, 5 (2017), S. 861–879

PyTrx: a Python-based monoscopic terrestrial photogrammetry toolset for glaciology. Frontiers in Earth Science 8:21, doi:10.3389/feart.2020.00021