340 likes | 508 Vues
This project developed an interactive search program for 3D geometric objects, allowing users to find similar objects based on feedback. The system is iterative, leveraging user input to refine search results. The project involved building a database of 3D colored models, performing data preprocessing, and designing tests with real users. The effectiveness of the search algorithms was evaluated against different configurations, with a focus on utilizing feature vectors derived from geometric and color characteristics. Conclusions indicate varied success across configurations, emphasizing the importance of user interaction.
E N D
Israel Institute of Technology Technion Faculty of Electrical Engineering Project A 044167 Summer 2001
3D Geometric Objects Search • Project team: • Lyakas Alexander 307666883 • Instructor: Dr. Sigal Ar
• Given a collection (database) of objects • Choose a search object • Find objects that are similar to the search object • The search is iterative and interactive • A user marks some objects as ‘GOOD’ or ‘BAD’ • The search program tries to refine the search by considering the user’s feedback The Main Idea
Starting the Project • The following components available from Iterative and Interactive Search for Objects by Moty Golan & Oren Kerem based on Similarity Between Three-Dimensional Objects – An Iterative and Interactive Approach by Michael Elad, Ayellet Tal, Sigal Ar. • Two databases: 3D colorless models & 2D images • A search program
Project Requirements • Test the approach with 3D colored models • Improve the search program • Design & perform system tests with real users
Project Requirements - Cont. • Build a database of 3D colored models • Gather 3D colored objects from the WWW • Perform preprocessing calculations, i.e. present each object in a way that will enable searching
Project Requirements - Cont. • Improve the search program • Adding a new database must not influence the search program’s code • Add features needed for testing
Project Requirements - Cont. • Test the system with real users • Design the tests • Perform tests with volunteers • Draw conclusions
• To calculate feature vectors we used moments of different orders on • colored points in 3D • colored normals in 3D • colorless points in 3D • colorless normals in 3D Working with Objects • Each object is presented as a numerical vector, AKA ‘feature vector’
• Consider two objects represented as feature vectors: • We can compare them using the (square of) standard Euclidean distance: • By adding weights and a bias value we can refine the distance function: Comparing Objects
Data Preprocessing • Convert the objects to the format convenient to be sampled • Perform sampling • Correct normals directions • Normalize rotation and scale • Create icons for all objects • Calculate features vectors
• The sampling workflow: • Choose a triangle to sample Sampling • Before sampling each object is presented as colored triangular mesh Ensure uniform sampling • Sample a point, normal and color from the chosen triangle • Do this as many times as needed (10,000 in our case)
• The pqr-th moment (of a 3D object) is defined as: • We approximate moments as: • Feature vector of level 3 in ‘colorless 3D’ look like: Calculating features • The order of the moment is p+q+r
• The object-based solution introduces the DBLINK class • Database-specific information is stored inside DBLINK objects only • One DBLINK object for each database – stored on disk The Search Program • The extendibility requirement – adding new database must not influence the search program code
• Automatic Screen Shooting • Before search refinements – with user’s ‘GOOD/BAD’ markings • When the new results are displayed Saving Test Sessions Results • Manual Screen Shooting
Testing the system • Several volunteers that had no previous knowledge about how the system works • Tests were done for several test objects • For each test object – all search configurations were tried • The testers gave feedback on the search results
• In most cases the search converged • not always with good search results… • ‘Normals’, level 4 worked good • but slow… • So should the colors be considered? • … Conclusions • Searching for objects having a ‘family’ was successful with most configurations • No search configuration worked well for all objects
The End • See the project book for many skipped details