Method and apparatus for context-based video quality assessment
View Patent ↗Because neighboring frames may affect how a current frame is perceived, we examine different neighborhoods of the current frame and select a neighborhood that impacts the perceived temporal distortion (i.e., when frames are viewed continuously) of the current frame most significantly. Based on spatial distortion (i.e., when a frame is viewed independently of other frames in a video sequence) of frames in the selected neighborhood, we can estimate initial temporal distortion. To refine the initial temporal distortion, we also consider the distribution of distortion in the selected neighborhood, for example, the distance between the current frame and a closest frame with large distortion, or whether distortion occurs in consecutive frames.
1. A method for estimating visual quality of a video sequence, comprising:
determining a neighborhood of a current frame of the video sequence from a plurality of neighborhoods that include the current frame;
determining initial temporal distortion for the current frame responsive to spatial distortion for frames in the determined neighborhood;
determining a distribution of the spatial distortion for the frames in the determined neighborhood;
determining temporal distortion for the current frame responsive to the initial temporal distortion and the distribution; and
determining the visual quality of the video sequence responsive to the temporal distortion for the current frame.
2. The method of claim 1 , wherein the determined neighborhood has a highest density of frames with large distortion among the plurality of neighborhoods and has most impact on the temporal distortion of the current frame among the plurality of neighborhoods.
3. The method of claim 1 , wherein the neighborhood of the current frame is determined based on a sliding window approach.
4. The method of claim 1 , wherein the determining the distribution comprises:
determining a distance between the current frame and a closest frame with large distortion in the determined neighborhood, wherein the temporal distortion is determined responsive to the initial temporal distortion and the distance.
5. The method of claim 4 , wherein the temporal distortion dt(n) for the current frame is determined as:
dt ( n )= d t,0 ( n )/dist( n ),
wherein d t,0 (n) is the initial temporal distortion for the current frame, and dist(n) is the determined distance.
6. The method of claim 1 , wherein the determining the distribution comprises:
determining a mean, minimum, or maximum of distances between frames with contiguous large distortion in the determined neighborhood, wherein the temporal distortion is determined responsive to the determined mean, minimum, or maximum distance.
7. The method of claim 1 , wherein the determining the distribution comprises:
determining a mean, minimum, or maximum of distances between contiguous distortion bursts in the determined neighborhood, wherein the temporal distortion is determined responsive to the determined mean, minimum, or maximum distance.
8. The method of claim 1 , further comprising:
performing at least one of monitoring quality of a bitstream, adjusting the bitstream in response to the determined visual quality, creating a new bitstream based on the determined visual quality, adjusting parameters of a distribution network used to transmit the bitstream, determining whether to keep the bitstream based on the determined visual quality, and choosing an error concealment mode at a decoder.
9. An apparatus for estimating visual quality of a video sequence, comprising at least a memory and one or more processors, the one or more processors configured to:
determine a neighborhood of a current frame of the video sequence from a plurality of neighborhoods that include the current frame;
determine initial temporal distortion for the current frame responsive to spatial distortion for frames in the determined neighborhood;
determine a distribution of the spatial distortion for the frames in the determined neighborhood;
determine temporal distortion for the current frame responsive to the initial temporal distortion and the distribution; and
determine the visual quality of the video sequence responsive to the temporal distortion for the current frame.
10. The apparatus of claim 9 , wherein the determined neighborhood has a highest density of frames with large distortion among the plurality of neighborhoods and has most impact on the temporal distortion of the current frame among the plurality of neighborhoods.
11. The apparatus of claim 9 , wherein the one or more processors are configured to determine the neighborhood of the current frame based on a sliding window approach.
12. The apparatus of claim 9 , wherein the one or more processors are configured to determine a distance between the current frame and a closest frame with large distortion in the determined neighborhood, and to determine the temporal distortion responsive to the initial temporal distortion and the distance.
13. The apparatus of claim 12 , wherein the one or more processors are configured to determine the temporal distortion dt(n) for the current frame as:
dt ( n )× d t,0 ( n )/dist( n ),
wherein d t,0 (n) is the initial temporal distortion for the current frame, and dist(n) is the determined distance.
14. The apparatus of claim 9 , wherein the one or more processors are configured to determine a mean, minimum, or maximum of distances between frames with contiguous large distortion in the determined neighborhood, and wherein the temporal distortion estimator is configured to determine the temporal distortion responsive to the determined mean, minimum, or maximum distance.
15. The apparatus of claim 9 , wherein the one or more processors are configured to determine a mean, minimum, or maximum of distances between contiguous distortion bursts in the determined neighborhood, and to determine the temporal distortion responsive to the determined mean, minimum, or maximum distance.
16. The apparatus of claim 9 , wherein the one or more processors are further
configured to perform at least one of monitoring quality of a bitstream, adjusting the bitstream in response to the determined visual quality, creating a new bitstream based on the determined visual quality, adjusting parameters of a distribution network used to transmit the bitstream, determining whether to keep the bitstream based on the determined visual quality, and choosing an error concealment mode at a decoder.
17. A non-transitory computer readable storage medium having stored thereon instructions for estimating visual quality of a video sequence, the estimating comprising:
determining a neighborhood of a current frame of the video sequence from a plurality of neighborhoods that include the current frame;
determining initial temporal distortion for the current frame responsive to spatial distortion for frames in the determined neighborhood;
determining a distribution of the spatial distortion for the frames in the determined neighborhood;
determining temporal distortion for the current frame responsive to the initial temporal distortion and the distribution; and
determining the visual quality of the video sequence responsive to the temporal distortion for the current frame.