Warning: Creating default object from empty value in /home/izonedesign/blog.bathdigital.com/wp-content/themes/canvas/functions/admin-hooks.php on line 160

Autostitch

I was introduced to Matt Brown by Guy McCusker, who works at the University of Bath. Autostitch has become my favourite-by-far photo app on the iPhone and when Guy told me it had been developed here in the city, I went in search of more information..

Q: First, please introduce yourself

I’m Matt Brown, a Lecturer at the University of Bath and a founder of Cloudburst Research, a startup specialising in Computer Vision on mobile devices. I first heard about Computer Vision when I was an undergraduate at Cambridge, then moved to Canada to do a PhD in the area. I started doing some research in object recognition and panorama stitching, and out of that came Autostitch.

Q: What is Autostitch?

Autostitch automatically combines photographs into seamless panoramas, using object recognition techniques to recognise overlaps and align the images. It’s based on research I did as a PhD student at the University of British Columbia (UBC). Later, I teamed up with Professor David Lowe and Dr Bob Hearn (also from UBC) to form a startup company called Cloudburst Research, which is now developing and improving the software for iPhone, iPad, Android etc.

Q: How does it differ from other products that do a similar thing?

Autostitch is a very flexible approach to panorama stitching as it can work with photos from any source and in any arrangement. Other panorama apps have a constrained capture process, such as a single sweep in the case of Apple’s panorama stitcher. They also tend to use lower quality and often blurry video footage, whereas Autostitch works with high resolution stills. This approach, combined with advanced algorithms for alignment and blending, gives us the highest quality on the App Store and results that are comparable with professional desktop software.

Richmond panorama

Q: Without getting too scientific (!) can you describe how it does what it does so well?

A key part of the approach is the use of Scale Invariant Feature Transform (SIFT) features to find matches between the images. This approach was developed by one of our co-founders, Professor David Lowe, and has been revolutionary in the field of Computer Vision. We are then able to implement the remaining stages of a professional panorama stitching pipeline, including bundle adjustment for optimal alignment of all the images, and multi-bland blending for high quality output. Dr Bob Hearn provided a lot of expertise to allow us to implement this efficiently and beautifully on iOS.

Q: What next for Autostitch?

We’re working on a new Android version, as well as several improvements to our iOS versions for iPhone and iPad. Our team has grown to 5 people, and we’re looking forward to the future of Computer Vision and Photography on mobile.

Q: What’s your favourite panorama taken with Autostitch?

That’s a difficult question! I spent a lot of time in the mountains when I was in BC and some of the most amazing panoramas I’ve shot were taken there. But now that Autostitch is always in my pocket it’s much easier to share panoramic views from anywhere… I stitched a really great panorama inside the stadium at the London Olympics last summer, for example.

Autostitch is available on the app store here.

About Mike

Mike Ellis is co-director of Bath consultancy Thirty8 Digital, BathCamp founder and Creative Director of Bath Digital Festival. In past lives he has been Head of Web at The Science Museum, Digital Strategist at Eduserv, Production Manager at Waterstone's Online and - once - a picture framer. At a push, he'll bang on about mobile technologies, social media, ubiquitous computing and innovation but ultimately he believes that shiny technology pales into insignificance next to the content it allows people to share. Mike lives in Bath with his wife and two sons.

Subscribe

Subscribe to our e-mail newsletter to receive updates.

No comments yet.

Leave a Reply