Street Sweeper: Detecting and Removing Cars in Street View Images
Wei-Ta Chu, Ying-Chieh Chao, and Yi-Sheng Chang
Multimedia Computing Laboratory
Dept. of Computer Science and Information Engineering
National Chung Cheng University
1. Introduction
To protect individualsˇ¦ and companiesˇ¦ privacy that may be leaked from street view images, we present a system that automatically detects cars and removes them as if they had never been there. Although street view service providers have made efforts on blurring human faces and license plates, we argue that the remaining features, such as license numbers and phone numbers printed on car bodies could still leak privacy. Given a sequence of street view images, this system first detects cars by the deformable part model, and then determines foreground seeds and background seeds fed to a GrabCut image segmentation module. After removing detected cars, an exemplar-based inpainting method is developed with special designs of filling priority determination and road structure propagation. Hierarchical texture propagation and randomized texture propagation are integrated to implement the inpainting process, so that aesthetically pleasing inpainting results as well as privacy protection can be accomplished.
2. Demonstration
2.1 Datasets
We manually captured Google Street View images along the same street from eight different places to form eight different datasets. Each dataset includes thirty spatially consecutive images, and different datasets have significantly different road situations.
ID |
Type |
Loc. |
Properties |
1 |
Highway |
Asia |
Traffic marks, crash barrier, fewer cars |
2 |
Highway |
Asia |
Traffic marks, crash barrier, fewer cars |
3 |
Highway |
USA |
Traffic marks, crash barrier, more cars |
4 |
Residential area |
USA |
Trees, buildings, parking cars, tree shadow |
5 |
Residential area |
USA |
Tree, buildings, parking cars, narrow lanes |
6 |
Tunnel |
Asia |
Dusky light |
7 |
Inside city |
Asia |
Complex traffic marks, buildings, intersection of roads |
8 |
Inside city |
Asia |
Complex traffic marks, buildings, trees |
2.2 Original image sequences and inpainting results (Please click thumbnails to see details.)
| Sequence 1 |
Original sequence |
 |
Inpainted sequence |
 |
| Sequence 2 |
Original sequence |
 |
Inpainted sequence |
 |
| Sequence 3 |
Original sequence |
 |
Inpainted sequence |
 |
| Sequence 4 |
Original sequence |
 |
Inpainted sequence |
 |
| Sequence 5 |
Original sequence |
 |
Inpainted sequence |
 |
| Sequence 6 |
Original sequence |
 |
Inpainted sequence |
 |
| Sequence 7 |
Original sequence |
 |
Inpainted sequence |
 |
| Sequence 8 |
Original sequence |
 |
Inpainted sequence |
 |
Any problem please contact
.
Last Updated:
April 16, 2014