Soccer Video Object Detection via Skeleton Pruning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting soccer players and balls in videos, especially in broadcast videos, suffer from errors such as false alarms and artifacts, due to the visual distinctiveness of the playing field and players' uniforms, which complicates accurate detection and tracking.
Innovation Solution
A shape analysis-based approach is employed to identify players and the ball by performing Euclidean distance transforms, skeleton pruning, and reverse Euclidean distance transforms to extract skeletons and remove artifacts, primarily caused by playing field lines, from roughly extracted foregrounds obtained through color segmentation and connected component analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If color segmentation is used to detect players and ball in soccer videos, then detection speed is improved, but detection accuracy deteriorates due to false alarms and artifacts from field lines
Solution Approach 1:
The patent divides the detection process into multiple stages: initial color-based segmentation to identify candidate regions, followed by shape analysis segmentation to separate true objects from artifacts. Euclidean distance transform is applied to segment the foreground into skeleton structures, enabling distinction between players/ball and field lines based on structural characteristics
Solution Approach 2:
The patent transforms the detection parameters from simple color-based identification to shape-based parameters including skeleton structure, compactness, and aspect ratio. This parameter transformation allows the system to maintain fast color-based initial detection while adding shape-based filtering to eliminate false alarms and artifacts
2Measurement precision
If morphological operations are applied to remove artifacts, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent replaces traditional morphological operations with Euclidean distance transform and skeleton-based processing. This substitution maintains artifact removal effectiveness while reducing computational complexity and processing time, as the distance transform efficiently identifies and removes linear artifacts like field lines without requiring multiple iterative morphological steps
Data Source
AI summary
A method is disclosed for detecting and locating players in soccer video frames without errors caused by artifacts by a shape analysis-based approach to identify the players and the ball from roughly extracted foregrounds obtained by color segmentation and connected component analysis, by performing a Euclidean distance transform to extract skeletons for every foreground blob, by performing a shape analysis to remove false alarms (non-players and non-ball), and then by performing skeleton pruning and a reverse Euclidean distance transform to cut-off the artifacts primarily caused by playing field lines.