Abstract
![CDATA[Shadows have a significant effect on the performance of many computer vision tasks, such as object tracking, action recognition, and structure health monitoring. In many object detection systems, shadows are often misclassified as parts of the moving objects or independent moving objects. As a result, the performance of these subsequent higher-level tasks is adversely affected. This paper presents a novel region-based method for detecting moving shadows by exploiting a new feature of pixel-geometry direction combined with the pixel-gradient magnitude. The new feature can be directly extracted from the given frame without prior knowledge about the scene or object properties. A major advantage of using such features for shadow classification is the ability to solve most of the problems associated with shadow detection in videos. Experimental results show that the proposed method is computationally faster and has higher detection rates and discrimination rates when compared to three well-known methods.]]
Original language | English |
---|---|
Title of host publication | 2016 International Conference on Digital Image Computing: Techniques and Applications (DICTA), Gold Coast, Australia, 30 November-02 December 2016 |
Publisher | IEEE |
Pages | 559-564 |
Number of pages | 6 |
ISBN (Print) | 9781509028962 |
DOIs | |
Publication status | Published - 2016 |
Event | DICTA (Conference) - Duration: 30 Nov 2016 → … |
Conference
Conference | DICTA (Conference) |
---|---|
Period | 30/11/16 → … |
Keywords
- image processing
- pattern recognition systems
- shadow detection