Coherent Label Placement for 3D Exploded View

2018 
The use of labels in images represents the basics of visual object presentations that we are all familiar with. However, few know that automatic label placement in 2D or 3D space belongs to the set of NP-complete and NP-hard problems. While state-of-the-art algorithms such as hedgehog labeling already produce incredible coherent results in real-time interactive applications, they were only designed for static and non-deformable objects. Therefore, their performance decreases when combined with the dynamic and model-deforming the 3D model presentation techniques such as exploded diagrams a.k.a. exploded views, which present the structure of 3D model by "exploding" their parts. We propose an extension of hedgehog labeling to work with exploded views by introducing clustering of model exploded parts and their labels. Clustered hedgehog labeling uses the explosion information to separate 3D space into sections belonging to individual label clusters, each running hedgehog labeling instances solely on the cluster members. The evaluation of the proposed solution and its Textplosion implementation was done by running a usability study enhanced with eye-tracking on a group of volunteers, where improvement of the original algorithm was detected. The need for 3D test models for experimentation resulted in the creation of a 3D Labeling dataset to be shared with the community in an attempt to fill the void of a missing standardized dataset for 3D labeling algorithms.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    0
    Citations
    NaN
    KQI
    []
    Baidu
    map