More iPhone Apps Using Computer Vision and Image Processing Technology

Since our previous post, Mobile Vision: iPhone Apps Employ Computer Vision and Image Processing Techs, more iPhone applications using computer vision and image processing technology have come out. Here we review two more apps.

  1. SnapTell: Tell the price, store, or online website of a CD/DVD/Book/Video Game with a photographed cover image. This is an image retrieval/recognition problem.
  2. CubeCheater: Solve your puzzle cube with six pictures taken for each face of the cube. It uses image segmentation and blob detection technology.

SnapTell is an image recognition based iPhone application. You can take a picture of the cover of any DVD/CD/Book/Video Game, send it to the SnapTell server, and you will receive the information for the product. The core technology for this application is image matching. According to SnapTell, the algorithm for image matching is called “Accumulated Signed Gradient” (ASG). The technical details are not disclosed, but it seems the so-called “accumulated signed gradient” is a kind of modified version of “Scale-invariant feature transform” (SIFT), where each detected feature point is described as the local gradient histogram of its neighborhood patch.

SnapTell describes the performance of their algorithm as follows,

Our robust matching engine can handle pictures taken in real life conditions that may have lighting artifacts, focus/motion blur, perspective distortion and partial coverage.

Show here are actual images that were sent in by users in a recent campaign that were matched 100% successfully by our image matching engine.

This advertisement is pretty aggressive from the technology point of view. It would be interesting to get more insights of the technology. 

Another iPhone app, CubeCheater, solves your favorite 80’s puzzle cube quickly. You only need to take pictures for each face of your cube, and CubeCheater will show you how to solve it step by step. Let’s watch a video to see how it works.

From the pictures you take, the algorithm first detects the edges of the picture (edge detection), and then recognizes each small cube by color and edge. The Kociemba algorithm is then used to find a solution quickly. The following snapshots illustrate the steps. Image processing and computer vision algorithm make this application pretty handy.

In summary, iPhone cameras are increasingly coupled with computer vision technologies for interesting applications. Have you come up with one?

1 Comment

Other Links to this Post

  1. How things work: Amazon Flow App Algorithm « the Serious Computer Vision Blog — January 29, 2012 @ 9:49 pm

RSS feed for comments on this post. TrackBack URI

Leave a comment