Ehsan Khoramshahi

PhD, Computer Science (University of Helsinki),

Current academic role

Guest Editor special edition in MDPI remote-sensing: “Advances in Deep Learning Approaches: UAV Data Analysis“

Former roles

1- Finnish AI region (FAIR) project manager from Metropolia University of Applied Sciences (MUAS) side.

2- Principal Lecturer in AI (MUAS).

3- Senior Research Scientist, Finnish Geospatial Research Institute (FGI), National Land-Survey of Finland (NLS).

4- Guest Lecturer in Photogrammetry.

5- Research Scientist, University of Helsinki.

6- Member of DroneFinland research group and DroneKnowledge project.

Former academic roles

1- Guest Editor special edition in MDPI remote-sensing: “Novel Applications of UAV Imagery for Forest Science“.

2- Guest Editor special edition in MDPI remote-sensing: “Remote Sensing for Spatial Information Extraction and Process“.

E-mail: Ehsan.khoramshahi (at) [photogrammetry.fi]; ehsan.khoramshahi (at) [visionium.fi]

Cite this web page: https://photogrammetry.fi/?page_id=22

The Network of Experts

In MUAS, I worked under supervision of Dr. Erkki Räsänen, Mr. Jari Olli, and Ms. Tiina Vuorio. I warmly appreciate their kind support during the period that I worked there.

In FGI I worked under the supervision of Prof. Dr. Eija Honkavaara, and Prof. Dr. Juha Hyyppä. I warmly appreciate their kind support for my research and developments.

My supervisors in University of Helsinki were Prof. Dr. Petri Myllymäki, and Prof. Dr. Arto Klami. These years were the best parts of my life. I warmly appreciate their support.

Recent publications

  1. Ehsan Khoramshahi, Somayeh Nezami, Petri Pellikka, Eija Honkavaara, Yuwei Chen, Ayman Habib, A Taxonomy of Sensors, Calibration and Computational Methods, and Applications of Mobile Mapping Systems: A Comprehensive Review, (Accepted to be published), April 2025, MDPI Remote-sensing.
  2. Xuanzhi Liu, Ehsan Khoramshahi, Ruochen Zhang, Chen Chen, Eetu Puttonen, Yuwei Chen, Using Direct-Georeferencing with a MEMS-LiDAR to Classify Traffic Signs, an Automatic Approach,  2024 5th International Conference on Smart Sensors and Application (ICSSA), Penang, Malaysia, 2024, pp. 1-6, doi: 10.1109/ICSSA62312.2024.10788657.
  3. Ehsan Khoramshahi, Photogrammetric Computations, Visionium Oy, Vol.I, Rel. 1, Oct. 2024.
  4. Mohammad Imangholiloo, Ville Luoma, Markus Holopainen, Mikko Vastaranta, Antti Mäkeläinen, Niko Koivumäki, Eija Honkavaara, Ehsan Khoramshahi, A New Approach for Feeding Multispectral Imagery into Convolutional Neural Networks Improved Classification of Seedlings, 2023, Remote-Sensing, journal article. My roles: supervising and advising the student, helping to design the AI method, editing the article.
  5. Ehsan Khoramshahi, Roope Näsi, Stefan Rua, Raquel A Oliveira, Axel Päivänsalo, Oiva Niemeläinen, Markku Niskanen, Eija Honkavaara, A Novel Deep Multi-Image Object Detection Approach for Detecting Alien Barleys in Oat Fields Using RGB UAV Images, 2023, Remote-Sensing, journal article. My roles: Experimental design, visualization, original writing, editing.
  6.  Jianxin Jia, Haibin Sun, Changhui Jiang, Kirsi Karila, Mika Karjalainen, Eero Ahokas, Ehsan Khoramshahi, Peilun Hu, Chen Chen, Tianru Xue, Tinghuai Wang, Yuwei Chen, Juha Hyyppä, Review on active and passive remote sensing techniques for road extraction, 2021, Remote-Sensing, journal article. My role: editor.
  7.  Ehsan Khoramshahi, Raquel A Oliveira, Niko Koivumäki, Eija Honkavaara, An image-based real-time georeferencing scheme for a UAV based on a new angular parametrization, 2020, Remote-Sensing, journal article. My roles: Experimental design, visualization, original writing, editing.
  8. Somayeh Nezami, Ehsan Khoramshahi, Olli Nevalainen, Ilkka Pölönen, Eija Honkavaara, Tree species classification of drone hyperspectral and RGB imagery with deep learning convolutional neural networks, 2020, Remote-Sensing, journal article. My roles: Experimental design, visualization, original writing, editing.
  9. Ehsan Khoramshahi, Mariana Batista Campos, Antonio Maria Garcia Tommaselli, Niko Vilijanen, Teemu Mielonen, Harri Kaartinen, Antero Kukko, Eija Honkavaara, Accurate calibration scheme for a multi-camera mobile mapping system, 2019, Remote-Sensing, journal article. My roles: Experimental design, visualization, original writing, editing.
  10. Ehsan Khoramshahi, Eija Hokavaara, Modelling and automated calibration of a general multi-projective camera, 2018, Photogrammetric Record, journal article. My roles: Experimental design, visualization, original writing, editing.
  11.  Jian Tang, Yuwei Chen, Antero Kukko, Harri Kaartinen, Anttoni Jaakkola, Ehsan Khoramshahi, Teemu Hakala, Juha Hyyppä, Markus Holopainen, Hannu Hyyppä, SLAM-aided stem mapping for forest inventory with small-footprint mobile LiDAR, 2015, Forests, journal article. My roles: Experimental design, editing.
  12. Ehsan Khoramshahi, Juha Hietaoja, Anna Valros, Jinhyeon Yun, Matti Pastell, Image quality assessment and outliers filtering in an image-based animal supervision system, 2015, International Journal of Agricultural and Environmental Information Systems (IJAEIS), journal article. My roles: Experimental design, visualization, original writing, editing.

Recent Developments

1-     ImageLab is my C++ based user interface that is coupled with  the various processing engines that I developed for my research. It has been developed with the goal of creating real-time image-processing algorithms.

     It includes data input modules, as well as processing, output and visualization modules.

List of ImageLab’s capabilities:

a.      A graphical user-interface that acts as a laboratory for algorithm design and comparison.

b.     Many input units developed to allow image inputs to feed into the processing units. Input units for video, matrix and binary inputs also have been developed.

c.      Fourier spectral analysis has been developed and optimized (FFT).

d.     Unsupervised clustering techniques such as :K-Means, Kohonen SOM, and hierarchical clustering are developed and added.

e.      Supervised classification by multi-layer Perceptron (feed-forward Neural network), as well as Logistic-regression (LR) have been added.

f.      Convex-optimization methods, such as gradient descent, Barzilai-Borwein, and conjugate gradients have been implemented.

g.     An optimized and parallel implementation of Matrix class has been developed. Efficient multiplication based on aligned data with SSE2/SSE3/parallelization has been added.

h.     Matrix factorization methods (QR with pivoting, LU, SVD based on Householder transformation) has been developed.

i.       An optimized version of Intel MKL has been integrated into my matrix implementation. The computational core has migrated from 32-bit to 64-bit architecture.

j.       Weighted Linear least-square solver has been added.

k.     Constraint weighted linear least-square solver has been added.

l.       Keypoint extractor (SIFT, SURF), and robust matching has been added.

m.    Pyramid-wise matching with projective kernel has been adde.

n.     …

2-     DAG  is my C++ based implementation of a general Bayesian network. It includes scoring functions for structure learning, and exact and approximate inference methods. (snapshots) (pdf) (source)

a.      Classes for nodes, clusters, and sep-sets have been implemented.

b.     Junction Tree Algorithm (JTA) has been implemented.

c.      Exact inference has been added.

d.     Maximum a posterior estimate (MAP) has been developed.

e.      Scoring functions (BIC,AIC) for structure learning have been added.

f.      Soft/hard evidence entry has been implemented.

g.     Graphical user interface has been added, now DAG acts independently and as a module inside ImageLab.

h.     …

3-     BundleCore is my C++ based implementation of a constraint-based Bundle-Adjustment routine. It includes the following parts:

a.      The Brown’s model for direct and inverse camera calibration modeling.

b.     Collinearity, intersection and resection solver based on weighted linear least-square.

c.      Co-planarity solver to estimate an Essential matrix and decompose it to a set of orientations and rotations.

d.     Exterior orientation estimator based on weighted linear least-square. Also, Constraint weighted non-linear least square.

e.      Network estimator for creating a massive reference map.

f.      Automatic coded target reading (T1-T3 customized targets for the calibration room).

g.     3D registration module.

h.     Multi-projective camera (MPC) sensor model.

i.       Single-frame camera Bundle Block-adjustment least square estimator.

j.       Multi-projective camera calibration routine.

k.   Sparse inner-constraint block bundle adjustment.

l.       A real-time flight direction estimator and GPU-based sparse point-cloud generator is currently under my research focus.

4-     BundleGUI is my basic GUI developed specifically for the Bundle-Core. It is able to load the same project, and be statically linked to the computational core. It is specifically designed to be used for the manual reading and labeling of an image’s points. It is planned to be merged with the ImageLab.

Mrs. Somayeh Nezami

Co-Founder and Research Scientist, Visionium Oy.

Mrs. Somayeh Nezami is a Geospatial Information Systems (GIS) specialist with software development skills. She has a M.Sc. degree in engineering. She has research experience in machine learning projects. She worked on a plant type recognition project based on remote-sensing at Finnish Geospatial Research Institute (FG) in 2017-2018.