Recognition of 3D objects with uniform surface reflectance
Název práce v češtině: | Recognition of 3D objects with uniform surface reflectance |
---|---|
Název v anglickém jazyce: | Recognition of 3D objects with uniform surface reflectance |
Akademický rok vypsání: | 2005/2006 |
Typ práce: | diplomová práce |
Jazyk práce: | angličtina |
Ústav: | Katedra softwaru a výuky informatiky (32-KSVI) |
Vedoucí / školitel: | ing. Martin Urban, PhD. |
Řešitel: | skrytý - zadáno a potvrzeno stud. odd. |
Datum přihlášení: | 10.01.2006 |
Datum zadání: | 10.01.2006 |
Datum a čas obhajoby: | 18.09.2007 00:00 |
Datum odevzdání elektronické podoby: | 18.09.2007 |
Datum proběhlé obhajoby: | 18.09.2007 |
Oponenti: | Tomáš Svoboda |
Zásady pro vypracování |
Methods based on matching of regions repeatably detected in a transformation-covariant manner ("distinguished regions") represent the state-of-the-art in visual recognition. Their core assumption is the ability to establish correspondence of regions that have the same 3D pre-image. The main limitation of the methodology is the restricted class of objects is applicable to - they fail for object with uniform surface reflectance.
Objectives: 1. Familiarize yourself with visual recognition methods based on distinguished region correspondences. Critically assess their strengths and limitations. 2. Propose invariant edges-based structure as distinguished regions suitable for objects with uniform surface reflectance. 3. Evaluate the proposed algorithm on a problem of suitable complexity such as recognition of car classification. |
Seznam odborné literatury |
[1] Paul Viola, Michael Jones. Robust Real-time Object Detection. In IJCV 2002.
[2] J.Šochman and J.Matas. WaldBoost. Learnig for Time Constrained Sequential Detection. In CVPR 2005. [3] J.Matas, O.Chlum, U.Martin and T.Pajda. Robust Wide Baseline Stereo From Maximally Stable Extremal Regions. In BMVC 2002. [4] S.Obdržálek and J.Matas. Sub-linear indexing for Large Scale Object Recognition. In BMVC 2005. [5] David G.Lowe. Object Recognition from Local Scale-Invariant Features. In ICCV 1999. |
Předběžná náplň práce |
Methods based on matching of regions repeatably detected in a transformation-covariant manner ("distinguished regions") represent the state-of-the-art in visual recognition. Their core assumption is the ability to establish correspondence of regions that have the same 3D pre-image. The main limitation of the methodology is the restricted class of objects is applicable to - they fail for object with uniform surface reflectance. Propose invariant edges-based structure as distinguished regions suitable for objects with uniform surface reflectance and evaluate the algorithm on a problem of suitable complexity such as recognition of car classification. |