Switching Filtering Algorithm with Image Preprocessing and Gesture Tracking

Denoising has always been a critical technology in the image processing community. Gaussian noise and impulse noise are the most common image noises. A single filtering algorithm cannot suppress these two noise conditions. Besides, existing hybrid flitering methods have high time complexity and are difficult to meet the real-time requirements. For this, we propose a filtering algorithm based on conditional switching, which establishes an adaptive median filtering subsystem and a fast guided filtering subsystem. It can automatically determine the main noise and select a suitable filtering subsystem for image denoising. To prove the effectiveness of our method, the proposed method is apply to a Kinect-based gesture tracking system. Experimental results show that the switching filtering algorithm can fulfill both accuracy and real-time performance requirements.