I’m thrilled to have joined the University at Buffalo’s Remote Sensing Lab (RSL) as a citizen scientist in the Citizen Continental-America Telescope Eclipse (CATE 2024) project. Our team is one of many across North America working together to capture a unique view of the Sun’s corona during the total solar eclipse on April 8th, 2024. More information on RS Lab at UB.
I’ve done research about change detection techniques and as it seems, deep-learning-based methods have the best outcome among the others. Here is the research about change detection using one of the most effective methods of identifying meaningful changes in very high-resolution (VHR) images. Abstract. Change detection is now one of the most important duties in remote sensing science. Over time it has been proven that deep-learning-based methods are one of the most effective ways to identify meaningful changes in a pair of co-registered images. In this project, we are using an end-to-end CD method based on an effective encoder-decoder architecture for binary change detection named UNet++, where change maps could be learned from scratch using a VHR dataset called LEVIR-CD+. UNet++ has been an improved version of UNet that uses a series of nested and dense skip connections, rather than only connections between encoder and decoder networks. This research has proven that the more we use nested outputs in the result, the better the results can get. click to download the research
This project aims to investigate and analyze the shape of the earth using observations of gravity acceleration. The natural and physical shape of the earth is the shape of the surface of the oceans under the effect of gravity and centrifugal force; This defined surface is an equipotential surface called the geoid. This surface is a complex and irregular surface due to the heterogeneous distribution of mass inside the earth and on its surface. At first glance, the earth can be considered a sphere with an average radius of 6371 km. But we know that the earth’s sphericity is not perfect, but the earth has a slight depression at the poles. Its equatorial radius is 6378 km and its polar radius is 6356 km. In the first section of this project, we estimate the radius of the earth at 12 points by assuming that the earth is spherical and by observing 12 stations. According to this assumption, we expect the radius to be equal at all points, but we get the opposite result. According to the latitudes of the stations, as we get closer to the lower latitude, the radius of the earth increases. (due to the elliptical shape of the earth and centrifugal force). In the second part, the acceleration of gravity is calculated from other interfaces. The above-mentioned project is available through the following link: Click to download
In many sciences, especially the science of geometric geodesy, we face problems that cannot be solved by analytical methods, or it is very difficult and impossible to solve such problems with analytical methods. Such problems include solving integrals, ordinary differential equations with the initial value (ODE), or calculating the numerical derivative of some functions. For this purpose, the use of numerical methods is considered. In this research project, the goal of solving the integral is to calculate the perimeter of the ellipse. To solve this integral, we first convert it into a differential equation and convert it into an initial value problem (IVP). After preparing the initial value problem using numerical methods, we proceed to solve the desired equation and obtain the set of points that are calculated as the best approximation using numerical methods. In this project, we will use the following three methods to numerically solve the problem of finding the perimeter of Earth’s ellipsoid: Euler’s method Heun’s method 4th order Runge-Kutta method You can see the full paper in the following link: Click to download
Augmented Reality (AR) and Virtual Reality (VR) are two terms that have been heard more in the past recent years and their use has been increasing in various industries. Maybe you have heard the name of virtual reality. Augmented reality, like virtual reality, has become more popular recently and many mobile applications have been created for it. For instance, in a cinema, augmented reality is used to make fictional films. In many exhibitions, users have utilized augmented reality to display their products and portfolios. AR has even made its way into some museums in recent years. In this article, an attempt has been made to investigate augmented reality from every aspect and its difference from virtual reality. Then, the relationship between augmented reality and the spatial location and display of things using AR is discussed. The full article is available to download via the following link. Click to download
There are numerous indices in Remote Sensing science that can help us find a particular target using satellite images. For example, Normalized Difference Vegetation Index (NDVI) is one of the simplest graphical indicators that is often used to analyze RS measurements and assess whether the target being observed contains green health vegetation or not. In this project, various statistics of a Landsat 8 image have been analyzed using the ENVI program. Then a number of different indices have been implemented on the image. In the end, the result of each index has been displayed. The mentioned article can be found in the following link: Click to download
In this project, we aimed to apply different filters and masks to a spatial subset of a satellite image. We use Landsat 8 meta-data file to obtain the image. Then some remote sensing indices have been created using the “band math” option in the ENVI program. After creating the desired indicators, we will use them to specify each of the complications and coverages. We need to separate the five desired land coverages using indicators and each of the bands. These five land covers are: Dense vegetation sparse vegetation Urban coverage Sea coverage Cloud coverage To obtain each of these land covers, we act as follows: Dense vegetation: we use the NDVI index and mask this image by giving the appropriate value. Dense vegetation cover: to mask this cover, we also use the NDVI index, which represents the vegetation cover. Urban coverage: To obtain urban coverage, we use the NDBI index and its appropriate values. Sea coverage: Blue areas are obtained using the NDWI index. Cloud coverage: Cloud area is also obtained by using one of the bands called Quality. Noise removal and edge extraction in panchromatic images In the first step, we loaded the panchromatic image from the QuickBird sensor into Envi classic software. The panchromatic image has the highest spatial resolution among the spectral bands. For this reason, it is considered as the best band for spatial analysis. After adding this band from the QuickBird satellite in the software, noises are observed on its surface. To eliminate the noise in the image, we must use low-pass filters. In this part of the exercise, we selected a low-pass Gaussian filter from the Enhancement section of ENVI classic to remove noises.…
Classification is a process during which different classes of an image are separated from each other. Usually, the classification result is displayed using different colors. There are two main types of classification. Supervised Classification Unsupervised Classification The difference between these two methods is in the type of input data and the classification algorithm. When we use supervised classification, we have to give a series of labels along with input images to the program. Using the given labels, the algorithm learns the characteristics of our class and by generalizing it, it can use the trained classification in the whole set of the input image. But the unsupervised classification does not need any labels. For several reasons, we may not be able to find suitable labels to train the algorithm. Unsupervised classification helps us in these kinds of situations. One of the main methods used in unsupervised classification by the algorithm is creating fake labels. In this project, we will examine different classification methods in Envi software. The following steps have been done in this project: Atmospheric correction of the image calculation of reflectance Unsupervised Classification by ISODATA method Unsupervised Classification by K-Means method Supervised Classification by Mahalanobis method Supervised Classification by Maximum likelihood method DEM extraction of the area to use in the decision-tree classification method Decision-tree classification method The complete version of the article is available through the following link: Click to download
The ellipsoid is the best proper geometric shape for the earth, which can show us a very close estimate of the real surface of the earth. Determining the position on the ellipse surface is one of the most used actions that is performed on the ellipse. In previous years, one of the main goals of creating geodetic networks and determining the position of the ellipse was to prepare a map. Currently, the preparation of maps in different dimensions and scales depends largely on geodetic coordinates. But with the passage of time, the need for these coordinates and determining the position on the ellipse increased more than before, and in more applications, the use of techniques for measuring the coordinates on the ellipse was considered. Some applications of geodetic coordinates include the determination of boundaries, urban management, engineering projects, hydrography, ecology, and assessment of natural disasters such as earthquakes, etc. In this project, we will use the technique of determining the position on the ellipse using different observations of length, vertical angle, horizontal angle, azimuth, baseline, and observations related to determining coordinates. The purpose of this project is to solve 3D geodetic networks and estimate the 3D Cartesian coordinates of the network points and determine their position on the ellipse using the non-linear least squares method. Click to download
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