Designing 4 routes on a topography using Civil 3D

A road is a linear way for the conveyance of traffic that mostly has an improved surface for use by vehicles. In this project, we designed 4 routes with different slopes along the way. We used the AASHTO standard for designing our roads. Contours of a specific area had been given to us. first, we transformed the contours to a surface using Civil 3D options. then designed four routes with different parameters. The full article is available through the following link: Click to download  

Threshold Target Detection of static objects using MATLAB

Target detection is one of the most important tasks in remote sensing of aerial images. It’s referred to finding a specific area or object in aerial images. In this project, I performed a threshold target detection technique in MATLAB. In order to do that we used different bands of the image. the reason to do that is because different bands have specific values for a particular object. Then we performed histogram equalization to enhance the quality of the image. after that, we used the enhanced image to find the number of containers in the picture. I used the bwlabel function in MATLAB to calculate the number of objects. You can download the full article to see the final pictures and results via the following link: Click to download  

Filtering an aerial image in spatial and frequency domains

Applying different filters to images can cause some changes in the quality of an image. For example, applying a gaussian filter to an image can make the image smooth. On the other hand, using a high-pass filter can sharpen our image. In this project, we have applied low-pass and high-pass filters to an image in the spatial and frequency domain. All of the research was done in MATLAB. Here you can see the results in each article. Applying filters to an image in the spatial domain Applying filters to an image in the frequency domain

Interior orientation of aerial images using Python

The purpose of interior orientation is to transform the ground coordinate system into a photographic coordinate system (reconstructing the internal geometry of the camera). This transformation is a two-dimensional to two-dimensional change, so we use the aforementioned transformations for this purpose. To perform this transformation and Obtain the conversion parameters requires control points such as control fiducial points. Then we need checkpoints to check the correctness and accuracy of the calculated parameters. Some of our control points will play a role in this action for us. The image coordinates of these points are provided with high accuracy by the camera manufacturer. In this project, I’ve written a Python code that can transform the coordinate of a chosen point from an aerial image to its ground coordinates using Affine, Conformal, Projective, and Polynomial transform. The pictures we used to capture the point are as follows:   The source code and data for the project is available via the following link: Click to download  

Transforming different coordinate systems and calculating the evolute of an ellipsoid

The evolute of an ellipse is the locus of all the centers of the ellipse curves. In general, this case is investigated for all curves, but in this text, we will specifically investigate the ellipse. The eccentricity parameter has a great effect on the width of the ellipse because this parameter can determine the shape and stretch of the ellipse. This parameter is expressed for all conical sections. If we want to specifically check the ellipse, its eccentricity must be a number between zero and one. The closer this value is to zero, the ellipse will become a circle, and the closer it is to one, the more the ellipse will be stretched. In the next step, a MATLAB function has been written to transform the geocentric Earth-Centered Earth-Fixed Cartesian to the geodetic coordinates using Hirvonen/Moritz numerical method. Click to download

Creating a 3D model in Agisoft Metashape

The purpose of this project is to build a 3D model of an area. in this project, we used approximately 60 aerial images taken from a particular area. In the end, an orthophoto mosaic has been obtained. furthermore, We can observe that a 3D model of the area is achieved. First step: Add images to Metashape software Second step: Aligning the images Third step: Creating a dense point cloud Fourth step: Creating the mesh surface Fifth step: making the orthophoto image The sixth step: making a DEM digital land model Seventh step: making Texture The full article is available through the following link: Click to download    

Creating a windows program using C# to view and edit database information

This project is the final sequence of a couple of researches that I did in the course of Principle of Database. In this part of the project, a windows application form has been designed to edit data in the database. The program can do the following: Add new companies to the Company table in the designed Linkedin-based database: The new companies added to the table can have the number of employees, the industry of the company, and spatial or geographic coordinates related to them. View the information of each company View the number of companies in every city The full article is available via the following link: Click to download  

Fourier Analysis using MATLAB

Fourier analysis is a method of defining periodic waveforms in terms of trigonometric functions. In this project, the goal is to analyze and calculate the Fourier coefficients. This project consists of different parts such as Fourier analysis, Audio signal analysis, and image analysis in the frequency domain. 1. Fourier analysis The following steps have been taken: calculated Fourier coefficients for 4 different signals displayed the plot of each Fourier signal calculated Fourier transform of the above-mentioned signals analyzed the difference between the Fourier transform of each signal, when various changes occur in its arguments The principle of Signal Sampling   2. Audio signals analysis In this part of the project, the analog sound of each number in a telephone has been given. Using these sounds, the audio related to each number is stored in a particular matrix. after that, various functions such as the FFT function have been applied to the matrix to take the data in the frequency domain. In the end, using the array of zeros, the audio of the sounds of a desired phone number has been created and saved in audio format. Here is the audio: 3. Image analysis In the last part of the project, two different images have been taken to the frequency domain, Using the FFT function in MATLAB.   The full research project is available via the following link: Click to download  

ISPRS Conference 2022 paper abstract

Title Effect of Transferring Pre-trained Weights on a Siamese Change Detection Network   Authors: Mohammad Aghdami-Nia, Reza Shah-Hosseini, Mohammad Salmani   Abstract Change Detection (CD) is one of the most crucial applications in remote sensing which identifies meaningful changes from bitemporal images taken from the same location. In this study, we investigate the influence of transferring pre-trained weights on the performance of a Siamese CD network using a benchmark dataset. For this purpose, an autoencoder with the same encoder architecture as in the Siamese model is trained on the whole dataset. Then, the encoder weights are transferred from the autoencoder and the Siamese model is trained in two modes. In the first mode, the transferred weights are frozen and only the decoder section of the Siamese models is trained while the second mode trains the whole model without freezing any part of the model. Moreover, the Siamese model is also trained without using the pre-trained weights to set the basis for comparisons. The results indicate that freezing the encoder results in a relatively lower performance but offers a considerable amount of temporal efficiency in the training phase. On the other hand, training the whole model after the weight transfer acquires the best result with an improvement of 12.43% in the Intersection over Union (IoU) metric.

Land use classification using AlexNet architecture

In this project, we used a UC Merced land use dataset that was available on Kaggle. The dataset was classified into 21 classes where there were agricultural lands, airplane pictures, baseball diamonds, beach,es and so on. Each class was stored in a separate folder. So the first step was to extract images from each folder. The images are 256x256x3. After extracting the images, we have to shuffle them in order to enhance the process of learning for the model.