WO2007007750A1 - 映像整合装置、方法、およびプログラム - Google Patents
映像整合装置、方法、およびプログラム Download PDFInfo
- Publication number
- WO2007007750A1 WO2007007750A1 PCT/JP2006/313765 JP2006313765W WO2007007750A1 WO 2007007750 A1 WO2007007750 A1 WO 2007007750A1 JP 2006313765 W JP2006313765 W JP 2006313765W WO 2007007750 A1 WO2007007750 A1 WO 2007007750A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- video
- matching
- frame
- deteriorated
- degraded
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
- H04N17/04—Diagnosis, testing or measuring for television systems or their details for receivers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
- H04N17/004—Diagnosis, testing or measuring for television systems or their details for digital television systems
Definitions
- Video alignment apparatus method, and program
- the present invention relates to a video processing technique, and more particularly to a video matching technique for temporally and spatially matching a reference video and a degraded video used for subjective quality evaluation of a video.
- An objective quality evaluation method that estimates the subjective quality using physical features of a video signal, which is the power of multiple video frames (still images) arranged along the time axis, includes a reference video and a degraded video. Is used to estimate a subjective quality evaluation value for a desired evaluation object.
- Degraded video refers to video that has deteriorated due to signal processing such as encoding of the reference video or network transmission, that is, loss in the evaluation target!
- the reference image and the deteriorated image can be appropriately matched when, for example, the number of pixels expands or contracts to an integral multiple, but only a few pixels can be obtained. If it is deformed! /, It can be fully supported.
- the video itself may be stretched or deformed by a few pixels, the reference video itself may be originally blurred, or the degree of deterioration may be large. There is also.
- Such video signals can be partially matched, but in many other areas they cannot be matched, and it may not be possible to optimize the matching between the reference video and the degraded video.
- the present invention is intended to solve such a problem, and is a video matching device capable of determining whether video matching is appropriate when obtaining an objective quality evaluation that an evaluator having specialized knowledge does not make an empirical judgment. It aims to provide a method, and a program. Means for solving the problem
- an image matching apparatus includes an arbitrary reference image having a plurality of reference image frame powers arranged along a time axis, and the reference image frame is arbitrarily selected.
- the deteriorated video frame force deteriorated due to the evaluation target of the input is input, and the degraded video is selected from a plurality of reference video frames having a time lag within a predetermined number of frames for each deteriorated video frame.
- Frame and spatial And a matching state detecting means for detecting a reference video frame in a matching state that is temporally matched, and a matching degree deriving means for deriving a matching degree indicating a matching degree between the reference video frame in the matching state and the deteriorated video frame.
- a matching information output means for outputting the reference video and the degraded video that are matched based on the matching state and the matching degree, and a degradation amount derivation for deriving a degradation amount indicating a degree of degradation between any two video frames.
- a first degree of degradation indicating the amount of degradation between the reference video frame and the degraded video frame in the matched state by controlling the degradation amount deriving unit by the matching degree deriving means, and from the matched state
- a second deterioration amount indicating a deterioration amount between the reference video frame and the deteriorated video frame that are shifted by a predetermined number of pixels is obtained, and based on a ratio between the first deterioration amount and the second deterioration amount. Calculate consistency It is what I did.
- the video matching method includes an arbitrary reference video having a plurality of reference video frame powers arranged along the time axis by the matching state detection means, and arbitrary evaluation of these reference video frames.
- the degraded video frame power degraded by the target is input, and the degraded video frame is selected from among the degraded video frames and a plurality of reference video frames having a time lag within a predetermined number of frames for each degraded video frame.
- a matching state detection step for detecting a reference video frame in a matching state spatially and temporally matched, and matching indicating a matching degree between the reference video frame in a matching state and a degraded video frame by a matching degree deriving unit.
- the reference image and the deteriorated image matched based on the matching state by the matching degree deriving step for deriving the degree of matching and the matching information output means A matching information output step for outputting a matching level, and a degradation amount deriving step for deriving a degradation amount indicating a degree of degradation between any two video frames by a degradation amount deriving unit.
- the program according to the present invention includes a plurality of reference video frame forces arranged along a time axis, and arbitrary reference videos having these reference video frames.
- Degraded video frame power that has deteriorated is input, and the reference video and degraded video are output spatially and temporally aligned and output to the computer of the video alignment device.
- a matching state detecting step for detecting a reference video frame in a matching state spatially and temporally matched with the deteriorated video frame from a plurality of reference video frames having a time lag within a predetermined number of frames, and matching
- the degree deriving means indicates the degree of matching between the reference video frame and the deteriorated video frame in the matching state.
- a matching degree deriving step for deriving the matching level, a matching information outputting step for outputting the reference video and the degraded video and the matching level based on the matching state by the matching information output means, and the degradation amount deriving unit A deterioration amount deriving step for deriving a deterioration amount indicating the degree of deterioration between the two video frames, and using the deterioration amount deriving step as a matching degree deriving step, A first deterioration amount indicating the amount of deterioration between the frame and a second deterioration amount indicating a deterioration amount between the reference video frame and the deteriorated video frame in a state where the matching state force is shifted by a predetermined number of pixels; And a step of calculating the degree of consistency based on the ratio of the first deterioration amount and the second deterioration amount.
- the degree of matching indicating the degree of matching is calculated based on the ratio of the second amount of deterioration indicating the amount of deterioration between the reference video frame and the deteriorated video frame in a state shifted by a predetermined number of pixels from the matching state.
- the degree of consistency can be calculated as an index indicating the degree of optimization of video matching.
- FIG. 1 is a block diagram showing a configuration of a video matching apparatus according to an embodiment of the present invention.
- FIG. 2 is a flowchart showing a video matching process of the video matching device according to the present embodiment.
- FIG. 3 is an explanatory view showing an example of a reference test video for format conversion confirmation.
- FIG. 4 is an explanatory view showing an example of a reference test image for position alignment and alignment range confirmation.
- FIG. 5 is an explanatory view showing an example of a reference test video for confirming a local deformation processable range.
- FIG. 6 is an explanatory diagram showing an example of a reference test video for confirming the luminance and color appearance distribution.
- FIG. 7 is an explanatory diagram showing an example of matching display timings.
- FIG. 8 is an explanatory diagram showing another example of matching display timing.
- FIG. 9 is an explanatory diagram showing another example of matching display timing.
- FIG. 10 is an explanatory diagram showing an example of macro synchronization processing in the time direction.
- FIG. 11 is an explanatory diagram showing an example of micro synchronization processing in the time direction.
- FIG. 12 is an explanatory view showing a calculation example of the deterioration amount and the inter-frame difference value in the matching state detection process.
- FIG. 13 is an explanatory diagram showing a change in matching characteristics depending on a matching state.
- FIG. 14 is an explanatory diagram showing changes in matching characteristics depending on the definition of the reference video.
- FIG. 15 is an explanatory diagram showing pixel shift.
- FIG. 16 is a block diagram showing a configuration of a video matching apparatus according to another embodiment of the present invention.
- FIG. 1 is a block diagram showing a configuration of a video matching apparatus according to an embodiment of the present invention.
- FIG. 1 is a block diagram showing a configuration of a video matching apparatus according to an embodiment of the present invention.
- the video matching device 100 is a signal processing device that performs arbitrary arithmetic processing on an input signal to obtain a desired output signal, and includes a plurality of video frames (still images) arranged along the time axis.
- An arbitrary reference video 1 that is powerful and a deteriorated video 2 that has deteriorated due to an arbitrary evaluation target such as sign signal processing or network transmission are input to the reference video 1, and the deteriorated video 2 is compared to the reference video 1.
- This is a device that outputs images that are spatially and temporally matched and various types of matching processing information related to matching processing.
- the present embodiment shows the amount of deterioration between the reference video frame and the deteriorated video frame when the reference video and the deteriorated video are in an alignment state that is most closely aligned in terms of space and time.
- the degree of matching indicating the degree of matching based on the ratio between the amount of deterioration and the second amount of deterioration indicating the amount of deterioration between the reference video frame and the deteriorated video frame in a state that is deviated by a predetermined number of pixels from the matching state. Is calculated and output.
- the video matching apparatus 100 includes a format conversion unit 10, a display timing matching unit 20, a video matching unit 30, a deterioration amount deriving unit 40, and a matching parameter deriving unit 50. It is realized by a signal processing circuit unit, an arithmetic processing unit, and further a storage unit.
- the arithmetic processing unit has a microprocessor such as a CPU or DSP and its peripheral circuits, and various functional units are realized by reading and executing the program from the microprocessor or the memory or storage unit of the peripheral circuit.
- the storage unit includes a storage device such as a hard disk memory, and stores various processing information used in the signal processing circuit unit and the arithmetic processing unit, video data of reference video and degraded video, and a program.
- the format conversion unit 10 includes a signal processing circuit and an arithmetic processing unit.
- the format conversion unit 10 has a function of converting the signal format of the degraded video 2 into the signal format of the reference video 1, and the degradation after format conversion obtained by the format conversion. It has a function to output video 2A.
- the display timing matching unit 20 includes a signal processing circuit and an arithmetic processing unit.
- the display timing matching unit 20 has a function of matching the number of video frames of the degraded video 2A after format conversion and the display timing with the reference video 1, and this matching. And a function to output the obtained degraded video 2B after timing matching.
- the video matching unit 30 includes a signal processing circuit and an arithmetic processing unit. The video matching unit 30 performs matching processing on the input reference video 1 and the deteriorated video 2B for each video frame, and further matches the reference video and the deteriorated video after the matching. Has a function to output various types of matching information related to video matching processing, such as the degree of matching.
- the video matching unit 30 is provided with macro synchronizing means 31, micro synchronizing means 32, matching state detecting means 33, matching degree deriving means 34, and matching information output means 35 as specific functional means.
- the macro synchronization means 31 derives a macro frame difference between the reference video 1 and the degraded video 2B by comparing transitions of predetermined video feature values extracted from the input reference video 1 and the degraded video 2B, respectively. It has a function.
- the micro-synchronizing means 32 includes a plurality of arbitrary degraded video frames from the reference video 1 and the degraded video 2B that have been macro-synchronized by the macro synchronizing means 31, and a plurality of temporal shifts within a predetermined number of frames with respect to the degraded video frames. Select the frame pair with the reference video frame, control the deterioration amount deriving unit 40 to acquire the deterioration amount of these frame pairs, and the reference video 1 based on the frame pair with the minimum deterioration amount among these frame pairs 1 And a function of deriving a micro frame difference between the degraded video 2B.
- the matching state detection means 33 for the reference image 1 and the deteriorated image 2B synchronized by the macro frame difference and the micro frame difference, for each deteriorated image frame, the time difference within the predetermined number of frames from the deteriorated image frame. It has a function of detecting a reference video frame in a matching state spatially and temporally aligned with the degraded video frame from among a plurality of reference video frames.
- the matching degree deriving means 34 has a function of deriving a matching degree indicating a matching degree between the reference video frame and the degraded video frame in the matching state. More specifically, the deterioration amount deriving unit 40 is controlled to shift the first deterioration amount indicating the deterioration amount between the reference video frame and the deteriorated video frame in the aligned state by a predetermined number of pixels from the aligned state. The second degradation amount indicating the degradation amount between the reference video frame and the degraded video frame is obtained, and the degree of matching is calculated based on the ratio of the first degradation amount and the second degradation amount. To do. At this time, the spatial feature (SI) of the reference video calculated based on the ITU-T P. 910 rules is multiplied by the above ratio to match. The degree may be calculated.
- SI spatial feature
- the matching information output means 35 has a function of outputting matching information related to the video matching processing, including the degree of matching for each deteriorated video frame.
- the degradation amount deriving unit 40 includes a signal processing circuit and an arithmetic processing unit, and has a function of deriving a signal-to-noise ratio between any two video frames as a degradation amount and any two video frames. A function of deriving an average value of pixel value differences between frames as an inter-frame difference value, and a function of outputting the deterioration amount and inter-frame difference value to the video matching unit 30.
- the degradation amount deriving unit 40 sets the signal-to-noise ratio between the reference video frame and the degraded video frame as the first degradation amount in response to an instruction from the matching degree deriving unit 34. calculate. Further, in accordance with an instruction from the degree-of-matching deriving means 34, the reference video frame in the aligned state and a plurality of images that are shifted from the aligned state in the horizontal direction, the vertical direction, and the vertical and horizontal directions by a predetermined number of pixels The average signal-to-noise ratio with the degraded video frame is calculated as the second degradation amount. Further, in accordance with an instruction from the matching state detection means 33, the signal-to-noise ratio between the reference video frame and the deteriorated video frame is calculated as the deterioration amount.
- the matching parameter deriving unit 50 includes a signal processing circuit and an arithmetic processing unit.
- the matching parameter deriving unit 50 compares a predetermined reference test video 3 with a degradation test video 4 in which the reference test video 3 has deteriorated due to an evaluation target, It has a function of deriving various alignment parameters used for video alignment processing.
- FIG. 2 is a flowchart showing the video matching process of the video matching device according to the present embodiment.
- reference video 1 and degraded video 2 include frame rate information or frame display time Z capture time information, and there are multiple reference video and degraded video before and after the frame to be processed. It is assumed that the processing proceeds in non-real time while accumulating frames for storage.
- the image matching apparatus 100 Prior to the alignment of the reference image 1 and the deteriorated image 2, the image matching apparatus 100 compares the reference test image 3 and the deterioration test image 4 by the alignment parameter deriving unit 50 and compares various alignment parameters. Deriving the meter (step 100).
- the reference test video 3 stored in advance in the video matching device 100 may be input to the evaluation target, and the degradation test video 4 as the output may be received.
- These reference test image 3 and deterioration test image 4 may be actual signals or data files composed of data strings indicating signal values.
- the degradation test video 4 is processed using the same processing method as the format conversion unit 10.
- the matching parameters are derived after 4 is in the same format as the reference test video 3.
- FIG. 3 is an example of a reference test video for format conversion confirmation.
- This reference test image 3A is a test image in which the brightness (including color) of the same pattern such as a checkered pattern is maximized and minimized.
- the video scaling ratio 5A of the reference test video 3 and the degradation test video 4 can be derived as a matching meter from the difference in pattern position and size. If the reference test video 3A is used, the luminance appearance range 5B in the reference test video 3 and the degradation test video 4 can be derived as a matching parameter from the maximum and minimum luminance values (including colors) of the reference video and the degradation video.
- FIG. 4 is an example of a reference test video for position alignment 'alignment range confirmation.
- This reference test image 3B is a test image with markers that pinpoint the positions at multiple points scattered over a wide area of the image. The surrounding pattern power can also be estimated.
- FIG. 5 is an example of a reference test video for confirming the local deformation processable range.
- This reference test image 3C is a test in which black and white frames are alternately inserted for every fixed number of pixels, for example, every pixel. This is the image.
- FIG. 6 is an example of a reference test video for checking the luminance / color appearance distribution.
- This reference test video 3D is an image (gray image) that repeats a gradual change in brightness within the screen from the minimum value (e.g. 0) to the maximum value (e.g. 255), or for each reference color that changes the color in the same way.
- This is a test image consisting of images (red image, green image, blue image).
- the luminance and color appearance distribution 5G (average value, variance, number of gradations) in the degradation test video 4 can be derived as a matching parameter.
- the format conversion unit 10 expands the video derived by the matching parameter deriving unit 50.
- the signal format of the degraded video 2 is converted based on the matching parameter 51 including the reduction ratio 5A and the luminance appearance range 5B, and the degraded video 2A after the format conversion is output (step 101).
- the reference video 1 is an uncompressed YUV format and the data format of the degraded video is an uncompressed RGB format, for example, Rec. ITU-R BT.601 "STUDIO ENCODING PARAME TERS OF DIGITAL TELEVISION FOR STANDARD 4: 3 AND WIDE-SCREEN 16: 9 ASPECT RATIOS ⁇ ⁇ is used to convert degraded video 2.
- the deteriorated video 2 is in a compressed format, it is converted to an uncompressed format in advance. If the aspect ratio is different, convert it so that it is the same. For example, if it can be simply calculated as an integer multiple, it is necessary to convert it to an arbitrary size. For example, “Digital image processing that works well” Chapter 7 “Creating images” Perform conversion to any size as in “Resolution Conversion”. Then, the degraded video 2A converted from the reference video is passed to the display timing matching unit 20.
- the video enlargement / reduction ratio derived by the matching parameter deriving unit 50 is used. Based on 5A, the image is converted for enlargement / reduction. Also brightness If the appearance range of a pixel is due to a difference in the standard based on the appearance range 5B, conversion that matches the appearance range shall be performed using linear transformation of luminance values.
- the display timing matching unit 20 performs processing such as frame interpolation in order to match the degraded video 2A format-converted by the format conversion unit 10 with the display timing of the reference video 1, and performs degraded video after timing matching.
- 2B is output (step 102). 7 to 9 show examples of display timing matching.
- the degraded video 2A is displayed using the video displayed at the display timing of the standard video 1 or the video that is temporally close to the display timing of the standard video 1. Interpolate.
- the reference video 1 itself does not have a constant frame rate
- the reference video 1 and the deteriorated video 2A are interpolated so as to be displayed at an accurate time interval different from these.
- the video matching unit 30 obtains the video enlargement / reduction ratio 5A, luminance appearance range 5B, spatial direction shift amount 5C, corresponding range 5D, and video deformation region range 5E obtained by the matching parameter derivation unit 50. Or, using the matching parameter 52 such as the appearance distribution 5G, transition between the three motion states, ie, the evaluation start state, the synchronization state, and the freeze state, while the reference image 1 and the deteriorated image after timing matching 2B video alignment processing is performed (steps 110 to 135).
- the matching parameter 52 such as the appearance distribution 5G
- the video matching unit 30 performs macro synchronization processing by the macro synchronization means 31 in order to achieve rough synchronization in the time direction (step 110).
- the macro synchronization means 31 calculates the video feature amount such as the average value of the luminance Z color difference ZRGB value for each frame or for a specific area of the reference video 1 and the degraded video 2B for a certain period of time. Transitions are compared, and the frame difference when the consistency between the two becomes the highest is derived as a macro time-direction shift, that is, a macro frame difference.
- FIG. 10 is an example of macro synchronization processing in the time direction. Specifically, as shown in Fig. 10, the time from the condition that the difference between each time series value is minimized by shifting the feature quantity such as the average luminance value in the time direction or the condition that the cross-correlation coefficient is maximized. Misalignment, ie macro Deriving the frame difference.
- FIG. 11 is an explanatory diagram showing an example of micro synchronization processing.
- the micro-synchronization means 32 is an arbitrary frame pair of the reference image 1 and the deteriorated image 2B after the macro synchronization, for example, the first frame of the deteriorated image 2B, and the reference image 1 corresponding thereto. 11 and the frame force obtained by the matching parameter deriving unit 50 as shown in FIG. 11 for each frame to be searched and each frame pair having a time difference within a predetermined number of frames.
- the degradation amount deriving unit 40 calculates a signal-to-noise ratio PSNR (Peak Signal to Noise Ratio) using Equation 1 described later as the degradation amount between these two frames.
- PSNR Peak Signal to Noise Ratio
- the micro synchronization means 32 selects the frame pair having the maximum signal-to-noise ratio PSNR calculated by the degradation amount deriving unit 40, that is, the smallest degradation amount, so that the reference video 1 and the degraded video are selected.
- the frame correspondence with the most consistent 2B is obtained, and this frame difference is derived as a microframe difference. This makes it possible to achieve micro alignment in the time direction.
- a synchronized state such a frame correspondence relationship between the reference video 1 and the degraded video 2B is referred to as a synchronized state.
- the frame pair force in the frame correspondence relationship between the reference video 1 and the degraded video 2B has a spatial positional relationship that minimizes the amount of degradation, specifically a positional relationship that maximizes the signal-to-noise ratio PSNR.
- the state is the alignment state in which the reference image 1 and the deteriorated image 2B are most aligned in time and space.
- the video matching unit 30 matches the reference video 1 and the reference video 1 that are in a synchronized frame correspondence until either the reference video 1 or the degraded video 2B after timing matching is the last video frame. From the degraded video 2B, the reference video target frame and the degraded video target frame are selected in sequence, and the reference video target frame and the degraded video target frame are the same as shown below.
- the processing loop in the initial state and the frozen state is started (steps 112 and 114).
- variables i and j indicate the reference video target frame number and the degraded video target frame number, respectively.
- Variables N and M indicate the reference video final frame number and the degraded video final frame number, respectively.
- Flag F1 indicates the synchronized state (0) and asynchronous state (1) of both images in the spatial direction.
- Flag F2 indicates the synchronized state (0), frame skip (1), and others (2: frame) in the time direction. Return state).
- the variable Count is the number of times the degraded video is frozen.
- the video matching unit 30 uses the matching state detecting unit 33 to obtain the amount of deviation in the time-space direction obtained by the matching parameter deriving unit 50.
- a degraded degraded video is generated by correcting degraded video 2B based on 5C, luminance and color information 5B, 5G, and the degradation amount and inter-frame difference value from the reference video are calculated by the alignment status detection process described later (step 120).
- the matching state detection means 33 passes the deteriorated image and the reference frame to be processed to the deterioration amount deriving unit 40, so that the deterioration amount deriving unit 40 and the inter-frame deterioration amount related to these frames are transmitted. Get the difference value.
- FIG. 12 is an explanatory diagram showing a calculation example of the deterioration amount and the inter-frame difference value in the matching state detection process.
- the degradation amount deriving unit 40 receives the degraded video target frame (j) of the degraded video 2B and the reference video target frame (i) of the reference video 1 corresponding thereto received from the matching state detection unit 33.
- the reference image target frame (i) force The reference image frame force having a time lag within a predetermined number of frames and the amount of deterioration are derived for each force frame pair. From these frame pairs, the frame pair with the minimum degradation amount and the matching state is selected, and the frame number of the reference video frame is derived.
- the degradation amount deriving unit 40 derives an inter-frame difference value between each of the immediately preceding frame and the reference video frame and the degraded video frame of the frame pair using Equation 2 described later. At this time, the deterioration amount deriving unit 40 calculates a difference value of the pixel values for each pixel of both video frames, and calculates an average difference value of pixels of the entire video frame as an inter-frame difference value. [0052]
- the video matching unit 30 determines whether or not the degraded video is frozen based on the inter-frame difference value of each target frame derived by the degradation amount deriving unit 40 by the matching state detection unit 33 (step 121). In other words, if the difference value for the reference image shows a certain value while that of the deteriorated image shows almost 0, it is determined that the deteriorated image has no change.
- step 121 If the degraded video is in a frozen state (step 121: YES), the video matching unit 30 sets the flag F1 to the asynchronous state (1) and sets the number of freezes to 1 (step 128). Then, the process proceeds to step 114 of the processing loop.
- the video matching unit 30 performs a matching degree calculation process described later by the matching degree deriving unit 34, and shows a matching degree indicating a matching degree between the reference video frame and the deteriorated video frame in the matching state.
- the matching information output means 35 matches the matching reference video or the matching deteriorated video composed of the reference video and the deteriorated video based on the matching state, the matching degree, the matching parameter (5A to 5G), and the synchronization information (F2). And consistency information such as the number of freezes (Count) are output (step 126). After resetting Count to 0 (step 127), the process proceeds to step 114.
- the video matching unit 30 executes the matching state detection process in the matching state detection means 33 in the same manner as in step 120 (step 130), it is determined whether or not the deteriorated video is frozen in the same manner as in step 121 (step 131). If the deteriorated video is in a frozen state (step 131: YES), the alignment state detection means 33 increments the number of freezes Count (+1) (step 135) and proceeds to step 114 of the processing loop.
- step 131: NO when the deteriorated video is not in the frozen state (step 131: NO), the video matching unit 30 performs macro synchronization processing similar to step 110 (step 132), and the matching information output means 35 performs the number of freezes. Count is output (step 133). Then, the flags F1 and Count are reset to 0 (step 134), and the process loops to step 114.
- step 114 the reference video target frame number i and the degraded video target frame number j are incremented, respectively, and if i and N or; j ⁇ M holds, the loop processing between step 112 and step 112 is repeated. Also, when both i and j reach N and M, the series of video alignment processing is completed.
- FIG. 13 is an explanatory diagram showing changes in matching characteristics depending on the matching state.
- FIG. 14 is an explanatory diagram showing changes in matching characteristics depending on the definition of the reference video.
- FIG. 15 is an explanatory diagram showing pixel shift.
- the degree of matching is derived using the following two features.
- the first feature is that, as shown in FIG. 13, when the matching is optimized, the signal-to-noise ratio between the frame of the reference image and the deteriorated image in the matching state and only one pixel from the matching state If the signal-to-noise ratio difference in the shifted state is large and matching is not optimized, the difference in the signal-to-noise ratio in the state shifted by one pixel from the matched state is small.
- the second feature is that when matching is optimized, the characteristic is the fineness of the reference image pattern as shown in Fig. 14. The higher the video, the more prominent the tendency is.
- the specific degree of matching calculation process by the degree-of-matching deriving means 34 is as follows. First, the inferiority in the matching state that is determined to be the most temporally and spatially matched between the reference video and the degraded video. As shown in the signal-to-noise ratio PSNR (Peak Signal to Noise Ratio) in Equation 1, the pixel difference value between the reference image and the deteriorated image with respect to the peak luminance (255), In other words, the logarithmic ratio with MSE shown in Equation 2 is calculated.
- PSNR Peak Signal to Noise Ratio
- Equation 2 Yin and Yout are reference images Z-degraded images, N is the number of pixels, and Y (X, y, i) is the pixel value at the position (x, y) of the i-th frame.
- the signal-to-noise ratio PSNR is calculated in the same manner for the state where the predetermined number of pixels is deviated from the matching state (near the matching state), and the average value is calculated as the deterioration amount (second deterioration (Quantity).
- the deterioration amount second deterioration (Quantity).
- eight states shifted from the aligned state by one pixel or 18 states shifted by two pixels are used.
- the SI value is the spatial feature defined in ITU-T P.910 (“Subjective video quality a ssessment methods for multimedia applications,” Aug. 1996.). Information) is calculated, and the degree of consistency is derived from Equation 3 using these values.
- ⁇ is a coefficient for normalizing the degree of matching.
- Consistency a X SI value X of reference image X (PSNRZ alignment near PSNR average value) (3)
- the amount of deterioration between the reference video frame and the deteriorated video frame when the reference video and the deteriorated video are in the aligned state that is the most consistent in terms of space and time.
- a reference image with a first degradation amount indicating Based on the ratio of the second degradation amount indicating the degradation amount between the frame and the degraded video frame, the matching degree indicating the degree of matching is calculated and output.
- the degree of consistency can be calculated as an index.
- the degree of matching is calculated by multiplying the spatial feature quantity (SI) of the reference image calculated based on the ITU-T P.910 standard by the ratio of the deterioration amount, Can be weighted by quantity.
- the degree of matching can be calculated stably and universally.
- the case where the signal-to-noise ratio PSNR is used as the deterioration amount has been described as an example.
- the present invention is not limited to this, and an average difference value of pixel values may be used. Even when using a signal-to-noise ratio, use a known signal-to-noise ratio other than PSNR.
- the video matching unit 30 and the degradation amount deriving unit 40 are realized by separate functional units.
- both of these functional units are both signal processing circuits.
- This is realized by a processing unit and an arithmetic processing unit. Therefore, as shown in FIG. 16, the deterioration amount deriving unit 40 may be realized by the same function unit as the video matching unit 30.
- various types of data can be exchanged between each functional means of the video matching unit 30 and the degradation amount deriving unit 40 very easily, and the required processing time can be shortened and one-dose resource can be reduced. it can.
- This also applies to the relationship between the format converter 10, the display timing matching unit 20, or the matching parameter deriving unit 50 and the video matching unit 30.
- the degraded video frame is fixed and a plurality of reference video frames and frame pairs are configured in the micro synchronization processing in step 111 has been described as an example. It is not limited. For example, fix the reference video frame and It is possible to execute the same micro-synchronization processing as described above in which a degraded video frame corresponding to this and a plurality of degraded video frames having a time lag within a predetermined number of frames with the degraded video frame may be configured.
- the force described as an example in which the micro synchronization means 32 and the alignment state detection means 33 are realized by separate functional means is not limited to this.
- both the micro synchronization means 32 and the matching state detection means 33 both control the deterioration amount deriving unit 40 to acquire the deterioration amounts of a plurality of frame pairs and select the frame pair having the smallest deterioration amount. It has a function. Therefore, realize these with the same functional means.
- the matched information output means 35 when the matched information output means 35 outputs the matched video, the reference video frame and the degraded video frame used for calculating the degree of matching are matched with the matched reference video and the matched degradation, respectively. It may be output as a video.
- the present invention is not limited to this.
- the video alignment processing may be performed using the reference video 1 after the interpolation.
- the matching parameter deriving unit 50 is provided in the present embodiment, the case where the matching parameter deriving unit 50 is provided has been described as an example. However, the matching parameter deriving unit 50 is not provided, and necessary parameters are derived within the process of FIG. Please do it.
- the signal-to-noise ratio PSNR is If the maximum deviation is derived as 5C in the spatial direction,
- the luminance display range 5B is derived as a matching parameter from the maximum and minimum luminance values (including colors) of the reference video frame and the deteriorated video frame.
- the brightness and color appearance distribution 5G average value, variance, number of gradations
- a matching parameter deriving unit 50 is provided to compare a predetermined test reference image with a test deteriorated image obtained from the test reference image according to the evaluation target, Since the various matching parameters used to match the reference video and the degraded video are derived, the degradation characteristics of the evaluation target are remarkable compared to the case of using the normal reference video and degraded video as described above. Can be derived. Furthermore, since the matching process is performed based on this matching parameter, appropriate video matching is performed even if the reference video or degraded video is blurred or the degradation caused by the evaluation target is large. be able to.
- the video matching device 100 can be used as an objective evaluation device.
- the video matching apparatus objectively evaluates the quality of video played on the receiving terminal in quality management of video distribution services and video communication services using an IP network such as the Internet.
- IP network such as the Internet.
- it is useful as a device that aligns the reference video and the degraded video temporally and spatially.
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
- Picture Signal Circuits (AREA)
Abstract
Description
Claims
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US11/921,209 US8094196B2 (en) | 2005-07-11 | 2006-07-11 | Video matching device, method, and program |
EP06768078.5A EP1903809B1 (en) | 2005-07-11 | 2006-07-11 | Video matching device, method, and program |
CA2611397A CA2611397C (en) | 2005-07-11 | 2006-07-11 | Video matching device, method, and program |
JP2007524656A JP4482031B2 (ja) | 2005-07-11 | 2006-07-11 | 映像整合装置、方法、およびプログラム |
CN2006800218895A CN101199211B (zh) | 2005-07-11 | 2006-07-11 | 视频匹配装置,方法以及程序 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2005201910 | 2005-07-11 | ||
JP2005-201910 | 2005-07-11 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2007007750A1 true WO2007007750A1 (ja) | 2007-01-18 |
Family
ID=37637141
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2006/313765 WO2007007750A1 (ja) | 2005-07-11 | 2006-07-11 | 映像整合装置、方法、およびプログラム |
Country Status (7)
Country | Link |
---|---|
US (1) | US8094196B2 (ja) |
EP (1) | EP1903809B1 (ja) |
JP (1) | JP4482031B2 (ja) |
KR (1) | KR100963614B1 (ja) |
CN (1) | CN101199211B (ja) |
CA (1) | CA2611397C (ja) |
WO (1) | WO2007007750A1 (ja) |
Families Citing this family (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP5172440B2 (ja) * | 2008-01-08 | 2013-03-27 | 日本電信電話株式会社 | 映像品質推定装置、方法およびプログラム |
JP2009260941A (ja) * | 2008-03-21 | 2009-11-05 | Nippon Telegr & Teleph Corp <Ntt> | 映像品質客観評価方法、映像品質客観評価装置、及びプログラム |
US9087125B2 (en) * | 2009-08-10 | 2015-07-21 | Pixel Forensics, Inc. | Robust video retrieval utilizing video data |
JP2011254240A (ja) * | 2010-06-01 | 2011-12-15 | Sony Corp | 画像処理装置、画像処理方法およびプログラム |
KR20120029690A (ko) * | 2010-09-17 | 2012-03-27 | 삼성전자주식회사 | 디스플레이 장치 및 그 영상 처리 방법 |
US8873642B2 (en) * | 2010-12-22 | 2014-10-28 | Verizon Patent And Licensing Inc. | Video content analysis methods and systems |
US9055279B2 (en) * | 2011-04-15 | 2015-06-09 | Tektronix, Inc. | System for natural language assessment of relative color quality |
CN102879660B (zh) * | 2011-07-11 | 2016-08-31 | 富泰华工业(深圳)有限公司 | 电子产品测试装置及方法 |
KR20130015179A (ko) * | 2011-08-02 | 2013-02-13 | 삼성디스플레이 주식회사 | 표시 장치 및 표시 장치 구동 방법 |
US9191553B2 (en) * | 2011-09-30 | 2015-11-17 | Intel Corporation | System, methods, and computer program products for multi-stream audio/visual synchronization |
US8730328B2 (en) | 2011-10-06 | 2014-05-20 | Qualcomm Incorporated | Frame buffer format detection |
US9659437B2 (en) | 2012-09-28 | 2017-05-23 | Bally Gaming, Inc. | System and method for cross platform persistent gaming sessions using a mobile device |
US9767642B2 (en) * | 2011-10-14 | 2017-09-19 | Bally Gaming, Inc. | System and method for cross platform persistent gaming sessions using a mobile device |
US9672688B2 (en) | 2011-10-14 | 2017-06-06 | Bally Gaming, Inc. | System and method for cross platform persistent gaming sessions using a mobile device |
US9626567B2 (en) | 2013-03-13 | 2017-04-18 | Visible Measures Corp. | Automated video campaign building |
JP6398979B2 (ja) * | 2013-08-23 | 2018-10-03 | 日本電気株式会社 | 映像処理装置、映像処理方法および映像処理プログラム |
KR102236561B1 (ko) * | 2014-12-31 | 2021-04-07 | 삼성디스플레이 주식회사 | 열화 보상 장치, 이를 포함하는 표시 장치 및 열화 보상 방법 |
US10674180B2 (en) * | 2015-02-13 | 2020-06-02 | Netflix, Inc. | Techniques for identifying errors introduced during encoding |
CN109076202B (zh) * | 2016-04-27 | 2021-03-16 | 索尼公司 | 图像投影设备、投影成像系统和校正方法 |
CN107465940B (zh) * | 2017-08-30 | 2019-10-25 | 苏州科达科技股份有限公司 | 视频对准方法、电子设备及存储介质 |
CN109996062B (zh) * | 2019-04-04 | 2020-08-11 | 广东省安心加科技有限公司 | 视频图像质量检测方法、装置、计算机设备和存储介质 |
KR102680344B1 (ko) | 2022-11-25 | 2024-07-02 | 주식회사 와이즈오토모티브 | 영상 시험 장치 및 방법 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2000078618A (ja) * | 1998-07-25 | 2000-03-14 | Tektronix Inc | ビデオ画像品質の評価方法 |
WO2001074072A1 (en) | 2000-03-27 | 2001-10-04 | Teranex, Inc. | Processing sequential video images to detect image motion among interlaced video fields or progressive video images |
JP2001326869A (ja) * | 2000-05-15 | 2001-11-22 | Kdd Media Will Corp | 映像信号・映像信号解析結果同時表示装置 |
JP2003047029A (ja) * | 2001-07-31 | 2003-02-14 | Shibasoku:Kk | 動きベクトルを用いる画像評価方法及び装置 |
JP2005064679A (ja) * | 2003-08-08 | 2005-03-10 | Kddi Corp | 画像特徴量抽出方式および画質評価方式 |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5446492A (en) * | 1993-01-19 | 1995-08-29 | Wolf; Stephen | Perception-based video quality measurement system |
EP0809409B1 (en) * | 1996-05-24 | 2001-08-29 | Matsushita Electric Industrial Co., Ltd. | Method and circuit to determine a noise value that corresponds to the noise in a signal |
DE69803830T2 (de) * | 1998-03-02 | 2002-09-12 | Koninklijke Kpn N.V., Groningen | Verfahren, Vorrichtung, ASIC und deren Benutzung zur objektiven Videoqualitätbewertung |
US6826294B1 (en) * | 1999-03-05 | 2004-11-30 | Koninklijke Philips Electronics N.V. | Block matching motion estimation using reduced precision clustered predictions |
CA2403665C (en) * | 2000-03-31 | 2007-12-04 | British Telecommunications Public Limited Company | Image processing |
JP2004080177A (ja) | 2002-08-13 | 2004-03-11 | Nippon Telegr & Teleph Corp <Ntt> | 映像品質評価装置、映像品質評価方法、映像品質評価プログラム及びそのプログラムを記録した記録媒体 |
CA2646805C (en) | 2003-08-22 | 2012-04-24 | Nippon Telegraph And Telephone Corporation | Video quality assessing apparatus, video quality assessing method, video quality assessing program, video aligning apparatus, video aligning method, and video aligning program |
-
2006
- 2006-07-11 US US11/921,209 patent/US8094196B2/en active Active
- 2006-07-11 CN CN2006800218895A patent/CN101199211B/zh active Active
- 2006-07-11 KR KR1020077029000A patent/KR100963614B1/ko active IP Right Grant
- 2006-07-11 CA CA2611397A patent/CA2611397C/en active Active
- 2006-07-11 WO PCT/JP2006/313765 patent/WO2007007750A1/ja active Application Filing
- 2006-07-11 JP JP2007524656A patent/JP4482031B2/ja active Active
- 2006-07-11 EP EP06768078.5A patent/EP1903809B1/en not_active Ceased
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2000078618A (ja) * | 1998-07-25 | 2000-03-14 | Tektronix Inc | ビデオ画像品質の評価方法 |
WO2001074072A1 (en) | 2000-03-27 | 2001-10-04 | Teranex, Inc. | Processing sequential video images to detect image motion among interlaced video fields or progressive video images |
JP2001326869A (ja) * | 2000-05-15 | 2001-11-22 | Kdd Media Will Corp | 映像信号・映像信号解析結果同時表示装置 |
JP2003047029A (ja) * | 2001-07-31 | 2003-02-14 | Shibasoku:Kk | 動きベクトルを用いる画像評価方法及び装置 |
JP2005064679A (ja) * | 2003-08-08 | 2005-03-10 | Kddi Corp | 画像特徴量抽出方式および画質評価方式 |
Non-Patent Citations (1)
Title |
---|
See also references of EP1903809A4 |
Also Published As
Publication number | Publication date |
---|---|
EP1903809A1 (en) | 2008-03-26 |
CA2611397A1 (en) | 2007-01-18 |
US8094196B2 (en) | 2012-01-10 |
EP1903809B1 (en) | 2014-03-19 |
CA2611397C (en) | 2010-12-21 |
US20090040311A1 (en) | 2009-02-12 |
JP4482031B2 (ja) | 2010-06-16 |
JPWO2007007750A1 (ja) | 2009-01-29 |
KR20080012354A (ko) | 2008-02-11 |
EP1903809A4 (en) | 2011-10-12 |
CN101199211A (zh) | 2008-06-11 |
CN101199211B (zh) | 2010-05-26 |
KR100963614B1 (ko) | 2010-06-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP4482031B2 (ja) | 映像整合装置、方法、およびプログラム | |
JP5347012B2 (ja) | 映像整合装置、映像整合方法及び映像整合プログラム | |
US20100271515A1 (en) | Image generation apparatus and image generation method | |
US8305489B2 (en) | Video conversion apparatus and method, and program | |
US8036479B2 (en) | Image processing apparatus and method, and storage medium for controlling gradation of moving images | |
US7412092B2 (en) | Image processing apparatus, image processing method, and program | |
US20120274845A1 (en) | Image processing device and method, and program | |
KR101140442B1 (ko) | 이미지 상태 정보의 정정 방법, 움직임 보상 이미지 처리방법, 및 이미지 상태 정보 정정기 | |
JP4449696B2 (ja) | 画像処理装置および画像処理方法、並びにプログラム | |
US8115863B2 (en) | Video de-interlacer using pixel trajectory | |
JP2008005183A (ja) | 映画素材に適した映像符号化方法および符号化装置 | |
CN100440982C (zh) | 判定视频信号的设备和方法 | |
KR101174589B1 (ko) | 지역 복잡도를 기초로 한 디인터레이싱 방법 및 이러한 방법을 수행하는 영상 처리 장치 | |
JP4102620B2 (ja) | 映像方式変換装置 | |
JP5300600B2 (ja) | 映像処理装置およびその方法 | |
JP5082933B2 (ja) | 動きベクトル検出回路、動きベクトル検出装置、及び集積回路 | |
JP4883029B2 (ja) | 動きベクトル検出回路、動きベクトル検出装置、及び集積回路 | |
JP2010251880A (ja) | 映像処理装置およびその制御方法 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 200680021889.5 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2007524656 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 11921209 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2006768078 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2611397 Country of ref document: CA |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1020077029000 Country of ref document: KR |
|
NENP | Non-entry into the national phase |
Ref country code: DE |