A Saliency Map Fusion Method for Community Images
A fusion method and image technology, applied in the field of computer vision
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0048] In this embodiment, the method is divided into a training phase and a testing phase.
[0049] Such as figure 1 As shown, there is a training set D in the training phase; the corresponding benchmark binary label G; there are M extraction methods.
[0050] Execute step 100, for the image I in D, apply m kinds of extraction methods to extract the saliency map of the training image. The extraction results of various methods are S={S 1 ,S 2 ,S 3 ,...,S i ,...,S M}, S i Denotes the saliency map extracted by the i-th method. Step 110 is executed, and the reference binary value corresponding to the image I is marked as G. S i Comparing with G, calculate the AUC value (the area under the ROC curve), the extraction method with a large AUC value shows that the performance is very good, sort the results of the extraction method, and get the sorting table T i . Execute step 120, according to the calculation method of step 100 and step 110, obtain the sorting list of the e...
Embodiment 2
[0064] Such as figure 2 As shown, there are four extraction methods used,
[0065] No. 1 is the FT method (Achanta, R., Hemami, S., Estrada, F. and Susstrunk, S. (2009) 'Frequency-tuned salient region detection', Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition (CVPR) , 20–26 June, Miami, FL, USA, pp.1597–1604.);
[0066] No. 2 is the SEG method (Rahtu, E., Kannala, J., Salo, M. and Heikkila, J. (2010) 'Segmenting salient objects from images and videos', The European Conference on Computer Vision (ECCV), 5–11 September , Crete, Greece, pp.366–379.);
[0067] No. 3 is the CB method (Jiang, H., Wang, J., Yuan, Z., Liu, T., Zheng, N. and Li, S. (2011) 'Automatic salient object segmentation based on context and shape prior' , The British Machine Vision Conference (BMVC), 29 August–2 September, Dundee, Scotland, pp.1–12.);
[0068] No. 4 is the RC method (Ming-Ming Cheng, Guo-Xin Zhang, Niloy J. Mitra, Xiaolei Huang, Shi-Min Hu. Global Co...
Embodiment 3
[0075] A social image has two parts of information: image information and label semantic information. During the training phase, prior knowledge of the ranking of extraction methods that perform well for an image extraction can be obtained. Processing flow such as image 3 As shown, the tag semantic information 301 of the social picture 300 is person and grass, and the image information of the social picture 300 is 302 . Four extraction methods are used to extract the saliency map of the social picture 300, and 311, 312, 313, and 314 in the extraction result 311 are the saliency maps obtained by the four extraction methods. Comparing the standard binary map 320 of the saliency domain in the extraction result 311, the ranking map 330 is obtained, and 331, 332, 333 and 334 in the ranking map 330 are obtained by using the FT method, the SEG method, the CB method and the RC method respectively According to the results sorted by AUC value, it can be seen that the larger the AUC v...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More