Projects

Salt Detection Using Segmentation of Seismic Image

(SEP,2018-DEC,2018)

In this project, I present state-of-the-art deep convolution neural network (DCNN) to segment seismic image for salt detection below the earth surface. Detection of salt location is very important for starting mining. Hence, a seismic image is used to detect the exact salt location under the earth surface. However, precisely detecting the exact location of salt deposits is very difficult. Therefore, professional seismic imaging still requires expert human interpretation of salt bodies. This leads to very subjective, highly variable renderings. Hence, to create the most accurate seismic images and 3D renderings, I need a robust algorithm that automatically and accurately identifies if a surface target is salt or not. Since the performance of DCNN is well-known and well-established for object recognition in image, DCNN is a very good choice for this particular problem. I successfully applied DCNN to a dataset of seismic images in which each pixel is labeled as salt or not.

WIFI Based Indoor Localization Using Decision Tree Algorithm

(SEP,2018-NOV,2018)

In this project, I developed an indoor localization system using state-of-the-art decision tree technique from wifi signal. For localization, GPS (Global positioning system) is widely used. However, the performance of GPS degrades significantly inside any building. Hence, new technology is needed to solve this problem. As we all know, every modern building is equipped with numerous numbers of wifi routers and every person carries a smartphone with an embedded wifi receiver. By measuring the received strength of wifi signal (RSS) from the smartphone, it is possible to locate the phone as well as the person. There exist three types of approaches to solving the problem known as triangulation, scene analysis, and proximity. My approach lies in scene analysis. I applied decision tree technique to a dataset contains 2D location information, RSS value recorded from a smartphone, IMU (Inertial measurement unit) sensor reading from a smartphone and a smartwatch.

Design of a Fuzzy-PID Controller for Controlling Quadrotor Attitude

(AUG,2017-DEC,2017)

Our problem was based upon the lateral movements, circular trajectory, and hovering of a quadrotor. The advantages of this fuzzy system is that it solves the problem of fixed gains, is cheaper to develop, covers a wide range of operating conditions are more readily customizable, and are robust in terms of uncertainty. The challenges we faced with this system is that is nonlinear, under actuated, low on board processing capability, low operation time and low efficiency in power consumption.

Structure Learning of Bayesian Networks Using Genetic Algorithms

(JAN,2018-MAY,2018)

In this project we investigate the problem of the search for the best Bayesian network structure, given a database of cases, using the genetic algorithm philosophy for searching among alternative structures. We start by assuming an ordering between the nodes of the network structures. This assumption is necessary to guarantee that the networks that are created by the genetic algorithms are legal Bayesian network structures. Next, we release the ordering assumption by using a “repair operator” which converts illegal structures into legal ones. We present empirical results and analyze them statistically. The best results are obtained with an elitist genetic algorithm that contains a local optimizer.

Moving Object Detection Using Background Subtraction from a Stationary Camera

(JAN,2018-MAY,2018)

Detection of moving objects in a video sequence is a difficult task and robust moving object detection in video frames for video surveillance applications is a challenging problem. Object detection is a fundamental step for automated video analysis in many vision and digital image processing applications. Object detection in a video is usually performed by object detectors or background subtraction techniques. Frequently, an object detector requires manual labeling, while background subtraction needs a training sequence. In this project proposes, we attempt to solve the problem of detecting moving object with fixed camera and identifying a non-moving background with a moving object of interest using the latter technique.

Google Sites

Report abuse