Analysis of attack features in a broadcast handball video with image processing

Abstract

This study proposes a semi-automatic method for analyzing broadcast handball game footage using image processing technology and explores the characteristics of offensive strategies based on the analysis. Utilizing a prior study that determines camera direction and captures player positions and movement trajectories, we semi-automatically extracted metrics such as the “number of attacks,” “ball retention rate,” “player positions,” and “player movement trajectories.” Player positions and movement trajectories were extracted directly from the video footage. We applied our method to a match from the 2019 Women’s Championship Tournament between Russia and China. The results revealed distinct offensive strategies: In Russia, where the scoring rate was higher, players utilized lateral and diagonal movements to evade defenders and secure shooting lanes. In China, where the scoring rate was lower, players had difficulty penetrating Russian defenders and were forced to shoot lone shots without a clear shooting path.