1. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Processing a time (OX axis)—frequency (OY axis) representation in terms of spectral blocks (N is the number of blocks). Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

2. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Visualization of the audio features of a documentary (upper plots) and a music video (lower plots). Except for the CP the OY axis of all patterns is frequency. The SCP has a lower frequency resolution. The OX axis for the SP, DSP, VDSP, SCP is related to the evolution over time. For the LFP the OX axis is periodicity. The CP is a correlation matrix. Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

3. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Action-based temporal segmentation (the OX axis is the temporal axis, vertical blue lines correspond to shot changes). “Hot-action” and “low-action” segments are indicated in red and green, respectively. Letters denote the processing steps as described in the text. Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

4. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Similarity-based contour search for all contours of the entire Corel collection (60,000 images). The contour of the first image in each row is the selected sample contour, the remaining images in each row contain the most similar contours. The percentage on the left denotes correct basic-level categorization for the first 99 similar images. Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

5. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Average contour feature vectors for each genre (see Sec. 3.4). Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

6. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Average audio feature vectors for each genre (“cont.” stands for contrast, and “pat.” for pattern, see Sec. 3.1). Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

7. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Average color-action feature vectors for each genre (see Secs. 3.2 and 3.3). Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

8. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Precision (P) against recall (R) for different runs and amounts of training data (increases along the curves from 10% to 70%; the encircled results are detailed in Table 3). Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

9. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Overall average Fscore and correct classification (CD¯) for all genres against the amount of training data. Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

10. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. SVM multiclass classification results, from left to right: average precision versus recall, overall average Fscore and correct classification (CD¯) for all genres against the amount of training data used. Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

11. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Average confusion matrix (50% training data, i.e., 105 sequences, 15 per genre; abbreviations: “anim.”—animated, “comm.”— commercials, “doc.”—documentaries, “mov.”—movies, “mus.”—music). Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017

12. Date of download: 9/18/2016 Copyright © 2016 SPIE. All rights reserved. Feature-based 3-D movie representation in a spherical coordinate system (inclination-θ, azimuth-φ, radius-r). Each movie from the data set is represented by a point with which we associate an image vignette. Views A to E are screenshots taken from different perspectives (the points of view used are shown in the chart). In views A to E, representative genres are annotated (a demo is available at Video 1, MOV, 33.2 MB http://imag.pub.ro/ bionescu/index_files/MovieGlobe.avi). [URL: http://dx.doi.org/10.1117/1.JEI.21.2.023017.1] Figure Legend: From: Video genre categorization and representation using audio-visual information J. Electron. Imaging. 2012;21(2):023017-1-023017-17. doi:10.1117/1.JEI.21.2.023017