BANGALORE, India — MindTree Ltd. has unveiled a real-time video stitching algorithm as part of a video surveillance product. The company claims its patent-pending algorithm can stitch video from up to four cameras in real time.
Stitched video from multiple cameras increases field of view, providing better situational awareness to the viewer. The algorithm requires no scene training and performs dynamic feature matching and blending. A scene overlap of at least 5 percent is required for the algorithm to work.
Video stitching is typically deployed to provide greater coverage in railway stations, airports, retail stores along with border and coastal surveillance points. Video stitching is also deployed in high-end automobiles to enhance passenger safety by increasing the driver's field of view.
Vinod Deshmukh, Mindtree's president and CEO of R&D services, said it is offering "a suite of video analytics algorithms that include algorithms for display enhancement, video content analysis and video search and summarization."
The video surveillance solution includes three components: video analytics software, intelligent video management software and platforms. The analytics suite consists of algorithms that address display enhancements, video content analysis and video search.
Reference designs are available for digital video recorders (4-channel analog DVR and 16 channel hybrid DVR) and camera encoders (2/4 channel).
MindTree said it will license each component and provide customer support.
—K.C. Krishnadas is site editor of TechOnline India
