Detection of circles in images captured by in-vehicle camera using the ADAS development platform

Ovaj rad se bavi metodama za detekciju krugova na slikama snimljenim kamerom montiranoj na prednjoj strani vozila, te implementacijom metoda na ugradbenu ADAS razvojnu platformu. Prvo je dana teorijska podloga za metode koje se koriste u ovom radu. Metode koje se u radu koriste za detekciju krugova su Hough-ova transformacija i RANSAC algoritam. U sljedećem koraku je izrađen koncept rješenja u Python programskom jeziku na osobnom računalu. Nakon toga rješenje je implementirao i optimizirano u C programskom jeziku na osobnom računalu. Optimizirano rješenje je zatim implementirano na ADAS razvojnu platformu također u C programskom jeziku. Na razvojnoj platformi je slijedila daljnja optimizacija rješenja u pogledu pravilne raspodjele poslova na dostupne procesore razvojne platforme. Nakon optimizacije rješenja na ugradbenoj platformi uspoređene su performanse dvije metode. Rezultati su pokazali da rješenje pomoću Hough-ove transformacije bolje obavlja zadatak detekcije krugova na slikama, ali je nešto sporije od onog gdje je za detekciju korišten RANSAC algoritam.This paper deals with methods for detecting circles in images captured by a camera mounted on the front of the vehicle and implementing methods on the built-in ADAS development platform. First, the theoretical basis for the methods used in this paper are given. The methods used in this paper to detect circles are the Hough transformation and the RANSAC algorithm. In the next step, the concept of the solution in the Python programming language on a personal computer is developed. After that, the solution was implemented and optimized in the C programming language on a personal computer. The optimized solution was then implemented on the ADAS development platform also in the C programming language. The solution was further optimized on development platform in terms of the proper distribution of work on the available processors of the development platform. After optimizing the solution on the development platform, the performance of the two methods was compared. The results showed that the solution using the Hough transformation performs the task of detecting circles in images better, but is somewhat slower than where the RANSAC algorithm was used for detection