Accurate and relative fast constellation detection

Description

Our algorithm successfully achieves accurate and relative fast constellation detection. Unlike regular template matching problems, constellations are scatter-based, so existing techniques such as SIFT detector or SURF detector cannot be applied.

We build the database based on the information we need for constellation detection; then we apply the proposed algorithm to detect star patterns we are interested in. This enables the sky gazers and amateur photographers to extract star patterns of interest conveniently from their photos.

Detection Result with and without scale filter. (a)Detection results without scale filter, a lot of incorrect constellation matches are generated; (b) Detection result with scale filter, only the matches with proper scale is preserved.

REFERENCE

[1] Liebe, C. “Pattern Recognition of star Constellations for Spacecraft Applications”, IEEE AES Systems Magazine, Jan 1993
[2] Jiang, M., etc. “A Novel Star Pattern Recognition Algorithm For Star Sensor” Proceedings of the Sixth International Conference on Machine Learning and Cybernetics, 19-22 August 2007
[3] Rehman, M., etc. “Single Star Identification and Attitude Determination in Tracking Mode” International Conference on Control, Automation and Systems 2008
[4] M. D. Pham, etc. “A Star Pattern Recognition Algorithm for Satellite Attitude Determination” IEEE Symposium on Industrial Electronics and Applications September 23-26, 2012,
[5] Stellarium, Star Mapping Software, Matthew Gates, Barry Gerdes
[6] The constellations. International Astronomical Union. 5 June 2015. Retrieved from http://www.iau.org/public/themes/constellations/
[7] The Night Sky – 88 Constellations. AstronomyOnline.org. 5 June 2015. Retrieved from http://astronomyonline.org/Observation/Constellations.asp?Cate=Obse
rvation&SubCate=MP07&SubCate2=MP0801

Automatic detection of beer labels

Refocusing Photos using Depth from Defocus