Video evaluation apparatus, spatio-temporal frequency analyzing apparatus, video evaluation method, spatio-temporal frequency analyzing method, video evaluation program, and spatio-temporal frequency analyzing program
An object is to provide a video evaluation apparatus and video evaluation method capable of appropriately evaluating a video consisting of a plurality of frame images, using an objective standard. The object is achieved as follows. A reference characteristic, which is a predetermined spatio-temporal frequency characteristic about a video, is stored, and a video characteristic analyzer 101 calculates a spatio-temporal frequency characteristic of a target video consisting of a plurality of frame images. Then a video evaluation determiner 102 calculates an evaluated value of the target video, based on a relative relation between the calculated spatio-temporal frequency characteristic and the stored reference characteristic. This enables appropriate evaluation of the video.
1. A video evaluation apparatus comprising:
a memory configured to store a reference characteristic which is a predetermined spatio-temporal frequency characteristic corresponding to what can be perceived by human eyes;
a processor configured to calculate a spatio-temporal frequency characteristic of a video comprising a plurality of images, and calculate an evaluated value of the video, so that the evaluated value increases with decreasing difference between the calculated spatio-temporal frequency characteristic and the stored reference characteristic, wherein
the evaluated value is calculated by using a weighting function with at least the following quantities:
a motion feature quantity V that represents a motion feature quantity,
a temporal frequency feature quantity C that represents a temporal frequency feature quantity, and
a difference quantity ΔR that represents a difference between frequency components of the video and frequency components corresponding to what can be perceived by human eyes.
2. The video evaluation apparatus according to claim 1 , wherein
the processor is configured to adjust a predetermined frequency component for N (N: an integer of not less than 1) images between any two images in the video, based on the evaluated value, and calculate an evaluated value of the images the signal frequency component of which was adjusted.
3. The video evaluation apparatus according to claim 1 , wherein
the processor is configured to calculate spatial frequencies of a plurality of images, convert components of the calculated spatial frequencies into data associated with a predetermined reference, effect a frequency transformation on a temporal change of at least one spatial frequency component out of the converted spatial frequency components to calculate spatio-temporal frequencies, and calculate data capable of describing the spatio-temporal frequency characteristic of the video using a high-frequency component and a low-frequency component out of the calculated spatio-temporal frequencies.
4. A video evaluation method comprising:
calculating a spatio-temporal frequency characteristic of a video comprising a plurality of images; and
calculating an evaluated value of the video, so that the evaluated value increases with decreasing difference between the calculated spatio-temporal frequency characteristic and the reference characteristic being a predetermined spatio-temporal frequency characteristic corresponding to what can be perceived by human eyes, wherein
the evaluated value is calculated by using a weighting function with at least the following quantities:
a motion feature quantity V that represents a motion feature quantity,
a temporal frequency feature quantity C that represents a temporal frequency feature quantity, and
a difference quantity ΔR that represents a difference between frequency components of the video and frequency components corresponding to what can be perceived by human eyes.
5. A non-transitory computer-readable medium including a video evaluation program, which when executed by a computer, causes the computer to perform a video evaluation method comprising:
storing a reference characteristic which is a predetermined spatio-temporal frequency characteristic corresponding to what can be perceived by human eyes;
calculating a spatio-temporal frequency characteristic of a video comprising a plurality of images; and
calculating an evaluated value of the video, so that the evaluated value increases with decreasing difference between the calculated spatio-temporal frequency characteristic and the stored reference characteristic, wherein
the evaluated value is calculated by using a weighting function with at least the following quantities:
a motion feature quantity V that represents a motion feature quantity,
a temporal frequency feature quantity C that represents a temporal frequency feature quantity, and
a difference quantity ΔR that represents a difference between frequency components of the video and frequency components corresponding to what can be perceived by human eyes.
6. The video evaluation apparatus according to claim 1 , wherein
the motion feature quantity V is calculated with components of a motion vector and a frame rate of the video,
the difference quantity ΔR is calculated by averaging differences of a plurality of predetermined frames of the video, and
the temporal frequency feature quantity C is calculated by using weighting function with an average horizontal frequency component and an average vertical frequency component of the video.
7. The video evaluation method according to claim 4 , wherein
the motion feature quantity V is calculated with components of a motion vector and a frame rate of the video,
the difference quantity ΔR is calculated by averaging differences of a plurality of predetermined frames of the video, and
the temporal frequency feature quantity C is computed by using weighting function with an average horizontal frequency component and an average vertical frequency component of the video.
8. The non-transitory computer-readable medium according to claim 5 , wherein
the motion feature quantity V is calculated with components of a motion vector and a frame rate of the video,
the difference quantity ΔR is calculated by averaging differences of a plurality of predetermined frames of the video, and
the temporal frequency feature quantity C is computed by using weighting function with an average horizontal frequency component and an average vertical frequency component of the video.