In my thesis I am analizing vehicles that are moving from different aspects based on video records. For the observation I use digital image processing methods and learning algorithms.
In the first part I write about basic image processing methods.
In the second part I categorise the learning algorithms by efficiency, speed and storage capacity.
In the third part I describe the functioning of the software. After writing about the steps for initializig the project environment I describe how I subtract irrelevant regions form the video frames. After this step I write about how I create the background picture. Based on the background picture I define the lanes where cars have to be detected. The next step is to detect moving object on the driving lanes by DBSCAN algorithm. After this I make clusters with K-means algorithm for separating cars from other objects. The next step is to calculate the speed of the cars and to detect if a car would like to go to an other lane. Finally I describe the details of the user interface.