Tracking Pemain Sepakbola Menggunakan Metode Kalman Filter Berbasis Two-Stages Hungarian Algorithm

Rumaksari, Atyanta Nika (2017) Tracking Pemain Sepakbola Menggunakan Metode Kalman Filter Berbasis Two-Stages Hungarian Algorithm. Masters thesis, Institut Teknologi Sepuluh Nopember.

[img]
Preview
Text
2215206001-Master-Thesis.pdf - Published Version

Download (3MB) | Preview

Abstract

Tracking atau pelacakan pemain dalam video sepakbola menjadi bagian penting dalam aplikasi cerdas yang berbasis interaksi manusia dan computer. Dimana dalam proses ini obyek diberikan label sesuai dengan identitasnya atau peran dalam pertandingan. Proses ini bertujuan untuk membantu mencari obyek (pemain atau bola) secara efisien. Kemudian hasil lokasi pergerakan obyek tersebut disimpan dalam media, agar dapat digunakan untuk merekonstruksi pola tersembunyi dari pergerakan obyek. Tantangan utama dalam penelitian tracking ini adalah ketika obyek yang diambil memiliki ukuran kecil, jumlahnya banyak dan memiliki pergerakan yang random, terdapat bayangan oleh karena sistem pencahayaan dan posisi pemain sering berhimpitan (oklusi) sehingga menyulitkan detektor mendeteksi pemain. Peneliti telah berhasil menghadapi tantangan tersebut dengan membuat sistem komprehensif gabungan dua proses, yaitu proses deteksi obyek dan proses tracking. Pada proses deteksi obyek peneliti menggunakan metode background subtraction dengan menambahkan filter operasi bitwise sebagai perekonstruksi obyek dan penghilang bayangan. Selanjutnya, pada proses tracking, metode Kalman filter diaplikasikan dengan menambahkan metode penugasan dual-Hungarian pada proses estimasi obyek dan pemulihan garis trayektori akibat dari oklusi. ================================================================================= Tracking of players becomes an important part of intelligent applications based on human and computer interactions on soccer video. In this process the object is labeled according to its identity or role in the match. This process aims to efficiently locate the players automatically. It is also being used for storing the location for semantic approach. Thus, it can use to reconstruct the hidden patterns of object movement. The main challenge in this tracking are the object taken has a small size, number is large and has a random movement, there are shadows because of lighting system and position of players often coincide (occlusion). Thus, it is difficult for detectors to detect the players. We have successfully faced these challenges by creating a comprehensive system of combined two processes, namely object detection and tracking process. In object detection process, we used background subtraction method by adding bitwise operation filter as object reconstruction and shadow removal. Furthermore, in the tracking process, the Kalman filter method was applied using dual-Hungarian assignment method on the object estimation process and the result of occluded trajectory line.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Ilmu olahraga, subtraksi background,multi-tracking, visi komputer, computer vision, machine learning, sport sciences, tracking, background subtraction
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms
Divisions: Faculty of Industrial Technology > Electrical Engineering > (S2) Master Theses
Depositing User: Atyanta Nika rumaksari
Date Deposited: 15 Aug 2017 03:25
Last Modified: 15 Aug 2017 03:25
URI: http://repository.its.ac.id/id/eprint/42267

Actions (login required)

View Item View Item