JP4870990B2 - Image analysis apparatus, visual stimulus risk determination program, and image analysis system - Google Patents

Description

  The present invention relates to a video analysis apparatus for analyzing a video displayed on a display and determining a risk level indicating a possibility of causing a viewer to cause a light-sensitive seizure by a visual stimulus by the video, and a visual stimulus risk determination program Is.

  Conventionally, as a video analysis apparatus for determining the possibility of causing a viewer to have a light-sensitive seizure by visual stimulation of a video, a so-called harding machine (The Harding FPA Flash and Pattern Analyzer (see Non-Patent Document 1, etc.)) What is called a so-called reference measuring instrument (see Non-Patent Document 2, etc.) is known. Harding machines are developed based on ITC (Information Technology & Computing Services) 2001 Guidelines (see Non-Patent Document 3, etc.). On the other hand, the reference measuring instrument has been developed so that it can be checked substantially in accordance with the criteria of “Guidelines on video techniques such as animation” (see Non-Patent Document 4, etc.).

“Harding FPA Home flash & pattern analyzer” Cambridge Research Systems Ltd. homepage [online], [searched January 6, 2006], Internet “http://www.hardingfpa.co.uk” Emoto, Otsuka, Yamaga, Uehara, Iguchi, Ito, Fujii: “Development of“ Reference Measuring Instruments ”for Guidelines for Animation Programs”, ITE Technical Report, 22, No. 40, pp. 1-6 (1998) Independent Television Commission, U.K. “ITC Guidance Note for Licensees on Flashing Images and Regular Patterns in Television”, 1994, 1998, 2001-2002. Japan Broadcasting Corporation, Federation of Private Broadcasting, “Guidelines on Video Techniques for Animation”, 1998; http://www.nhk.or.jp/strl/publica/dayori/dayori98-10/kaisetsu1-2-j.html

By the way, in February 2005, ITU-R (International Telecommunication Union-Radio communication Sector) established an ITU-R recommendation BT. 1702 (ITU-R Recommendation BT.1702, “Guidance for the reduction of protective epileptic seizures caused by television”, 2005). In this recommendation, although not specified in the main text, as in the case of ITC guidelines as an appendix, blinking with a blinking luminance difference of 20 cd / m ² or more and the area of the blinking area exceeding 25% on both sides is 1 second. If it occurs 3 times (6 times or more in the number of brightness changes), it is highly likely to cause a photosensitive seizure.

Therefore, in conventional devices such as so-called harding machines and reference measuring instruments, ITC guidelines and ITU-R recommendations BT. A process for determining whether or not the image is in accordance with 1702, that is, an image that is highly likely to cause a photosensitive seizure, is performed as follows. For example, as a method of this determination processing, there is a method of performing whether or not the luminance difference between pixels at the same coordinates (location) between adjacent frames is 20 cd / m ² or more over all video frames. is there. As another determination processing method, there is a method in which the area of a pixel having a luminance difference of 20 cd / m ² or more exceeds 25% of the screen. In a so-called harding machine, blinking of luminance flicker or the like is detected while performing motion compensation.

That is, as described above, the conventional apparatus performs the determination process based only on “whether or not the reference is exceeded”. For example, in the conventional apparatus, even a video with a relatively low risk of having a luminance difference of 20 cd / m ² and an area ratio of slightly exceeding 25% is detected as dangerous and an alarm is issued. However, a video with a relatively low degree of risk that slightly exceeds the standard tends to occur at a high frequency. That is, in the conventional apparatus, a video with a relatively low risk is frequently detected as dangerous, and an alarm is frequently output. For this reason, even a very dangerous image may give a recognition that it is not dangerous. In addition, the conventional apparatus has a problem that a very dangerous image such as a flashing of the entire screen is missed although the luminance difference is slightly less than ²⁰ cd / m ² .

  Accordingly, the present invention has been made to solve the above-described problems, and is a video analysis apparatus for determining the degree of risk of visual stimulation using a relatively safe image or a very dangerous image, and a visual stimulus. It is an object of the present invention to provide a risk determination program and a video analysis system.

  The present invention was created to achieve the above object, and first, the video analysis device according to claim 1 analyzes a video to be displayed on a display and emits light to a viewer by visual stimulation by the video. In order to determine the degree of risk indicating the possibility of causing a susceptibility seizure, luminance classification storage means, area classification storage means, risk determination criterion storage means, luminance difference calculation means, luminance classification means, and area calculation Means, area classification means, and risk determination means.

  In such a configuration, the video analysis apparatus stores the luminance classification as a criterion for classifying the pixels into a plurality of sets according to the luminance difference between the pixels by the luminance classification storing means, and by the area classification storing means. , Storing the area classification as a criterion for classifying pixels into a plurality of sets according to the size of the blinking area, and further, by the risk judgment criterion storage means, photosensitivity seizure according to the luminance classification and the blinking area The determination criteria for determining the risk level indicating the possibility of causing the error are stored.

  In this video analysis device, the luminance difference calculation means calculates the luminance difference between arbitrary pixels located at the same coordinates between an arbitrary target frame in the frame constituting the video and its adjacent frame. Subsequently, in the video analysis apparatus, the luminance classification unit classifies the pixels of the luminance difference calculated by the luminance difference calculation unit according to the luminance classification stored in the luminance classification storage unit. Although the adjacent frame is preferably a frame that is temporally adjacent to the target frame itself, the ITU-R recommendation BT. According to 1702, as long as it is a frame that can detect blinking that occurs approximately three times per second (six times or more in the number of luminance changes), the concept may include a plurality of frames.

  Next, the video analysis apparatus counts the number of pixels for each luminance classification classified by the luminance classification unit by the area calculating unit, calculates the counted value as a blinking area according to the luminance difference, and subsequently Then, the blinking area calculated by the area calculating means is classified by the area classifying means according to the area classification stored in the area classification storing means.

Then, the video analysis apparatus is configured to determine, for each luminance classification, a risk criterion by referring to the risk criterion storage unit according to the classification result of the luminance classifier and the area classifier by the risk determination unit. Ru determine Teisu the risk in multiple stages by. In other words, as described above, since the luminance classification and the area classification are each classified into a plurality of groups, the risk determination means determines the risk for each luminance classification based on a determination criterion using these classifications as a scale. Determine in multiple stages.

Then, the image analyzer, the crisis Kendo determining means, in the degree of risk is determined in multiple stages for each luminance classification, select and output the highest risk.

Further, the video analysis apparatus according to claim 2 is configured to include down-sampling means in addition to the structure according to claim 1 .

In such a configuration, the video analysis apparatus samples the pixels by thinning out the pixels by the downsampling means. For example, the downsampling means thins out by calculating an average value of blocks for each pixel of the target frame.
Then, the video analysis apparatus performs each process using the pixels down-sampled by the down-sampling means by the brightness difference calculating means, the brightness classifying means, and the area calculating means.

According to a third aspect of the present invention, there is provided a video analysis apparatus including the luminance conversion means in addition to the first aspect.

In this configuration, the video analysis device converts the digitized video signal of the video into a luminance value by the luminance conversion unit, and the luminance difference calculation unit converts the pixels after the luminance conversion by the luminance conversion unit. The difference in luminance value is calculated.

The visual stimulation risk determination program according to claim 4 is for analyzing a video to be displayed on a display and determining a risk indicating a possibility of causing a viewer to cause a photosensitive seizure by the visual stimulation by the video. In addition, the computer is configured to function as a luminance difference calculation unit, a luminance classification unit, an area calculation unit, an area classification unit, and a risk determination unit.

  In such a configuration, the visual stimulus risk determination program uses the luminance difference calculation means to determine the luminance difference between arbitrary pixels located at the same coordinates between an arbitrary target frame in the frame constituting the video and its adjacent frame. Then, the luminance classification means reads out the luminance classification from the luminance classification storage means for storing the luminance classification as a criterion for classifying the pixels into a plurality of sets according to the luminance difference between the pixels, According to the luminance classification, the pixels having the luminance difference calculated by the luminance difference calculating means are classified.

  Next, the visual stimulus risk determination program counts the number of pixels for each luminance classification classified by the luminance classification unit by the area calculation unit, and calculates the counted value as a blinking area according to the luminance difference, Subsequently, the area classification means reads the area classification from the area classification storage means for storing the area classification as a criterion for classifying the pixels into a plurality of sets according to the size of the blinking area, and according to the area classification The blinking area calculated by the area calculating means is classified.

Then, the visual stimulus risk determination program shows the possibility of causing a light-sensitive seizure according to the luminance classification and the blinking area according to the classification result of the luminance classification means and the area classification means for each luminance classification by the risk determination means. with reference to risk criteria storage means for storing a criterion for determining the risk, Ru determine Teisu the risk at multiple stages criterion for determining the risk.
Then, the visual stimulus risk determination program selects and outputs the highest risk level among the risk levels determined in multiple stages for each luminance classification by the risk level determination means.

According to a fifth aspect of the present invention, there is provided a video analysis system including an SDTV (Standard Definition TeleVision) video signal and an HDTV (High Definition TeleVision) by the video analysis apparatus according to any one of the first to third aspects. ) System video signal processing system for parallel processing, comprising a first video analysis device, a second video analysis device, and control means.

  In such a configuration, the video analysis system analyzes the SDTV video signal using the first video analysis device, and analyzes the HDTV video signal using the second video analysis device. In the video analysis system, the control means causes the analysis of the SDTV video signal by the first video analysis device and the analysis of the HDTV video signal by the second video analysis device to be processed in parallel. ing.

The present invention has the following excellent effects.
According to the invention described in claim 1 or claim 4 , the luminance classification and the area classification are stored in the luminance classification storage means according to the ITC 2001 version guidelines and the “Guidelines on video techniques such as animation”. By setting the area classification storage means in multiple stages, the risk level can be determined in multiple stages by the luminance classification and the area classification. Therefore, according to the present invention, it is possible to determine the degree of danger of visual stimulation by a relatively safe image or a very dangerous image.

Therefore, according to the present invention, excessive attention should be paid to a video with a relatively low risk (and a relatively high frequency of appearance) that has a luminance difference of 20 cd / m ² and slightly exceeds an area ratio of 25%. On the other hand, it can call attention to very dangerous images that have been missed.

And since the highest risk is output, it does not make a user misjudg the risk.

According to the second aspect of the present invention, since the amount of data to be processed can be reduced, the speed of video analysis can be increased.

According to the third aspect of the present invention, it is possible to cope with the case of using a video system in which the correspondence relationship between the input pixel value to the display and the output luminance is different.

According to the fifth aspect of the present invention, the HDTV video signal and the SDTV video signal can be processed in parallel, and the gamma characteristics of both video signals are simultaneously analyzed by each video analysis device. You can also.

  The best mode for carrying out the present invention will be described below with reference to the drawings. The present invention relates to a video analysis device, a visual stimulus risk determination program, and a video analysis system. Here, the configuration and operation of the video analysis device will be mainly described.

[Outline of video analyzer]
First, an outline of the video analysis apparatus 1 will be described. FIG. 1 is a block diagram illustrating the configuration of a video analysis apparatus according to an embodiment of the present invention.
This video analysis apparatus 1 analyzes a video to be displayed on a display and determines a risk level indicating a possibility of causing a viewer to cause a light-sensitive seizure by a visual stimulus by the video.

  The video analysis apparatus 1 includes an A / D conversion circuit 2, a luminance conversion circuit 3, a flash memory 4, a frame memory 5, an image conversion circuit 6, an image measurement circuit 7, and a determination criterion changing unit 8. ing. The flash memory 4 particularly stores a luminance LUT (Look Up Table) 41, a luminance determination threshold table 42, an area determination threshold table 43, and a risk determination table 44. In addition, the video analysis device 1 executes a visual stimulus risk determination process based on a command signal from the control circuit 9.

<A/D conversion circuit>
The A / D conversion circuit 2 inputs video (video signal) as an analog signal, converts the analog signal into a digital signal, and outputs the digital signal to the luminance conversion circuit 3 as a video signal.

<Luminance conversion circuit>
The luminance conversion circuit 3 receives the video signal from the A / D conversion circuit 2, converts the voltage of the video signal into the luminance of the screen for each frame while referring to the luminance LUT 41, and outputs it to the frame memory 5. is there. This processing of the luminance conversion circuit 3 is referred to as pixel value conversion processing.

  FIG. 2 is a block diagram illustrating a more detailed configuration of the luminance change circuit of FIG. The luminance conversion circuit 3 includes a frame memory read / write unit 31, a flash memory read / write unit 32, a data / address conversion unit 33, and a luminance value calculation unit 34 in order to perform pixel value conversion processing.

  The frame memory read / write unit 31 reads a video signal written in the frame memory 5 and passes the video signal to the data / address conversion unit 33 and the luminance value calculation unit 34. The frame memory read / write means 31 writes image luminance value data to the frame memory 5.

  The flash memory read / write unit 32 reads data of the luminance LUT 41 from the flash memory 4 and passes the data to the luminance value calculation unit 34. The flash memory read / write means 32 also writes various data to the flash memory 4.

  The data / address conversion means 33 reads the video signal read from the frame memory 5 by the frame memory read / write means 31 as voltage value data, and converts it into an address designating a coordinate position on a display (not shown).

  The luminance value calculating unit 34 refers to the luminance LUT 41, calculates a luminance value for each pixel of the address converted by the data / address converting unit 33, and passes the luminance value to the flash memory read / write unit 32. .

  Next, a processing flow of the luminance conversion circuit 3 having the above-described configuration will be described. FIG. 3 is a block diagram illustrating the flow. Note that FIG. 1 and FIG. 2 will be referred to as appropriate. In the luminance conversion circuit 3, when a video signal is received from the A / D conversion circuit 2, the video signal is temporarily written in the frame memory 5 by the frame memory read / write means 31 (capture / Sa1). Then, the frame memory read / write unit 31 reads the video signal captured in the frame memory 5 as voltage value data and passes it to the data / address conversion unit 33. The data / address conversion means 33 converts the voltage value data into an address designating a coordinate position on a display (not shown) (Sa2).

  The luminance value calculation unit 34 refers to the luminance LUT 41 and calculates a luminance value for each pixel at the address converted by the data / address conversion unit 33. At this time, the luminance value calculation unit 34 generates two types of non-negative integer values with reference to the luminance LUT 41 (Sa3 or Sa5) for all combinations (8 bits × 3) of luminance and color difference signals. This non-negative integer value corresponds to the video signal voltage converted into the screen brightness. The non-negative integer value here is 8 bits. Further, in the luminance LUT 41, compensation values for input and output are defined in order to perform gamma compensation so that the nonlinear relationship in the gamma characteristic becomes linear. For example, in the luminance LUT 41, compensation values for a combination pattern of all levels of 16 × 32 luminance and color difference signals are defined according to the coordinate values of the frame.

  Since the value of the luminance LUT 41 is different between SDTV and HDTV, all of these processes are four systems. Therefore, in this embodiment, the case of HDTV will be described. Also, the HDTV-A system (Sa3, Sa4) and the HDTV-B system (Sa5, Sa6) are represented by the difference between the two types of non-negative integer values, that is, the luminance LUT 41.

  Then, the luminance value calculation unit 34 causes the frame memory read / write unit 31 to write all the calculated pixel luminance value data into the frame memory 5 (Sa4 or Sa6). In addition, it is desirable to process in parallel about the case of both HDTV-A type | system | group (Sa3, Sa4) and HDTV-B type | system | group (Sa5, Sa6). In other words, the video analysis device 1 can process the visual stimulus risk determination processing in the case of using a video system in which the correspondence between the input pixel value to the display device (display) and the output luminance is different. Next, returning to FIG.

<Frame memory>
The frame memory 5 inputs and stores the video signal from the luminance conversion circuit 3 and outputs the video signal to the image conversion circuit 6 and the image measurement circuit 7. The frame memory 5 stores each frame in order in a plane (area) corresponding to one screen of a display (not shown), and outputs a video signal to the image conversion circuit 6 and the image measurement circuit 7 in units of frames. It has become.

<Image conversion circuit>
FIG. 4 is a block diagram of the image conversion circuit of FIG. The image conversion circuit 6 includes a frame memory read / write unit 61, a first downsampling unit 62 (first downsampling process), a second downsampling process 63 (second downsampling process), and a motion vector detecting unit 64 ( Motion vector search processing). The frame memory read / write means 61 reads / writes data from / to the frame memory 5.

<First downsampling means>
The first down-sampling means 62 causes the frame memory read / write means 61 to read out the screen luminance data of one system that has undergone the first down-sampling process, and for the screen luminance data, the average value of the block for each pixel of the target frame is calculated. Is to be calculated. For example, in the case of screen luminance data of 1080 × 1920 pixels (in the case of HDTV), the average value (8 bits rounded off) for each (3/20) × (3/20) pixel is calculated. Then, the first down-sampling means 62 sets the calculated average value to a value of 160 × 268 pixels in HDTV. In SDTV, the value is 72 × 96 pixels.

  Further, for the HDTV vertical pixels, one line for each of the upper and lower lines is discarded, and the subsequent processing may be performed with 288 × 160 pixels. For frame pixels, one line for each of the upper and lower lines may be discarded, and the subsequent processing may be performed with 288 × 160 pixels. Therefore, the screen brightness data is thinned out by the first downsampling means 62.

<Motion vector detection means>
The motion vector detecting means 64 performs block matching on the range of all pixels of ± 9 pixels in the previous frame and ± 9 pixels in the previous frame, with 8 × 8 pixels as one block for one system of signals that have undergone the first downsampling process. To search for a motion vector in units of blocks. It should be noted that this search is not performed for the other signal system, and it is desirable to use the same motion vector in the subsequent processing. Note that the motion vector detecting means 64 repeats from the previous frame and goes back 30 frames. Subsequently, this process will be described in more detail. Note that the direction of the sign of the motion vector is negative on the upper and left sides of the position on the previous frame.

  The motion vector detection means 64 first calculates a value obtained by multiplying each pixel value by 64 (6 bit shift), and obtains an absolute value of a difference between values obtained by subtracting the sum of each block from the calculated value. Then, the motion vector detection means 64 performs block matching by searching for a relative position where the sum of the absolute values is minimum.

  Then, the motion vector detecting means 64 has a relative position (distance) close to 0 when the sum is the same value, and a higher position and a left position when the absolute values of the relative positions are equal. Select in order. Here, in order to cover the search outside the video, it is desirable to provide an extended region in which 8 pixels are vertically and horizontally filled with a fixed value “100” in each frame. In this case, the number of pixels in each frame is 304 × 176 pixels in HDTV. Note that the SDTV has 112 × 88 pixels. At this time, the inter-frame motion vector obtained for the 8 × 8 pixel block including the extension region is 38 × 32 in HDTV. Note that the number of SDTVs is 14 × 11.

  Note that it is desirable not to search outside the extended region. Therefore, in this case, for example, the range of values that can be taken by the motion vector of the upper left block including the extended region is 0 to 9 in both horizontal and vertical directions, and the upper left of the real image region before expansion (after expansion) In the block of 8 pixels inside from the upper left to the right and the lower), the horizontal and vertical are 8-9.

<Second downsampling means>
The second down-sampling means 63 performs, for each pixel in the real image area (part excluding the extension area) in the image including the extension area for which the motion vector search by the motion vector detection means 64 has been completed, The average value (8 bits rounded off) of the four pixels of the pixel, the lower pixel, and the lower right pixel is obtained. Then, the second downsampling means 63 newly sets the obtained average value as the value of each pixel. Accordingly, each pixel in the actual image area (a part excluding the extended area) is thinned out to the value of each new pixel by the second downsampling means 63.

  However, if processing is performed every other pixel in both horizontal and vertical directions, the position after compensation by the motion vector is in increments of one pixel, so it is desirable not to perform pixel thinning after calculating the average value. It is desirable to buffer the image after calculating the average value up to 30 frames before.

  Next, the processing of the image conversion circuit 6 will be described in more detail. FIG. 5 shows a block diagram of the processing flow. When the frame memory read / write means 61 reads the screen brightness value data from the frame memory 5 (5 (1)) (Sb1), the first downsampling means 62 thins the screen brightness value data (first thinning / Sb2). The result is written in the frame memory 5 (5 (2)) by the frame memory read / write means 61 (Sb3).

  When the frame memory read / write unit 61 reads the result of the first decimation from the frame memory 5 (5 (2)), the second downsampling unit 64 further decimates (second decimation / Sb4), The result is written into the frame memory 5 (5 (3)) by the frame memory read / write means 61.

  When the frame memory read / write means 61 reads the result of the first decimation from the frame memory 5 (5 (2)), the motion vector detection means 64 detects the motion vector (Sb6), and the result is stored in the frame. The memory 5 (5 (4)) is written by the frame memory read / write means 61 (Sb7). The frame memory 5 (5 (4)) at this time is particularly called a motion vector FIFO (First-In First-Out) storage means. The frame memory 5 is a FIFO. Further, the frame memory 5 (5 (n)) means that it is composed of a plurality of regions and a plurality of chips in the frame memory 5. Next, the image measurement circuit 7 will be described.

<Image measurement circuit>
FIG. 6 shows a block diagram of the configuration of the image measurement circuit of FIG. The image measurement circuit 7 includes a luminance difference calculation unit 71, a luminance classification unit 72, an area calculation unit 73, an area classification unit 74, a risk determination unit 75, a frame memory read / write unit 76, and a flash memory read unit. 77. Among these, the frame memory read / write means 76 reads and writes the frame memory 5. The flash memory reading unit 77 reads from the flash memory 4. Hereinafter, other means will be described.

  The luminance difference calculating means 71 calculates a luminance difference between arbitrary pixels located at the same coordinates between an arbitrary target frame in a frame constituting an image and its adjacent frame.

  The luminance classification unit 72 classifies the pixels having the luminance difference calculated by the luminance difference calculation unit 71 according to the luminance determination threshold table 42 as the luminance classification stored in the flash memory 4.

  Here, the brightness determination threshold value table 42 will be described. The luminance determination threshold table 42 is used to divide the blinking luminance difference detected from the input video into 16 levels. For example, it can be expressed as follows.

unsigned char TH [16] = / * Threshold levels * /
{4,5,6,7,9,11,13,16,19,23,28,33,39,46,54,64}

  Then, the luminance classification unit 72 refers to such a luminance determination threshold table 42, and when blinking is detected, the number of pixels blinking at each level is referred to as G [i]. Count to the array.

<Area calculation means>
The area calculation means 73 counts the number of pixels for each luminance classification classified by the luminance classification means 72, and calculates the counted value as a blinking area corresponding to the luminance difference. Therefore, the blinking area is calculated for each of the 16 levels.

  The area classification unit 74 classifies the blinking area calculated by the area calculation unit 73 according to the risk determination table 43 as the area classification stored in the flash memory 4. Here, the area determination threshold value table 43 will be described.

  The area determination threshold value table 43 is data for classifying the blinking area detected from the input image into 32 levels. For example, it can be expressed as follows.

unsigned short AR [32] = / * Area Size * /
{90,98,108,119,131,144,157,172,188,206,226,248,272,299,328,360,394,
432,474,520,571,626,686,753,825,905,993,1089,1194,1310,1436,1575}

  Then, the area classification unit 74 determines which level the risk level corresponds to using the count result G [i] of each pixel number obtained by leveling the luminance value and this AR [].

  The risk determination means 75 refers to the risk determination table 44 according to the classification results of the luminance classification means 72 and the area classification means 74, determines the risk in multiple stages, and outputs the risk. Here, the risk determination table 44 will be described, and the processing of the risk determination means 75 will be described.

<Danger level judgment table>
This risk determination table 44 shows the determination criteria of the risk on the basis of the luminance difference and the blinking area. Next, an example is shown. Here, a specific example of a 16 × 32 luminance pattern is shown. In the row direction (vertical direction), the luminance difference classification is shown so that the degree of risk increases from the upper row to the lower row. In the column direction (horizontal direction), the blinking area classification is shown such that the degree of risk increases from the left column to the right column. In the description, all the rows are shown up to the 23rd column, and the subsequent columns are omitted.

<Specific Example of Risk Determination Table (Luminance Difference 16 Steps x Blink Area 32 Steps)>
<Definition formula>
unsigned char PR [16] [32] = / * Probability * /
<Luminance pattern data string>
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 2, 2, ...
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 4, ...
0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 2, 3, 3, 4, 4, 5, 5, 6, 6, 6, 7, 7, 8,…
0, 0, 0, 0, 0, 1, 2, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9,10,10,10,…
0, 0, 0, 1, 2, 3, 4, 5, 5, 6, 7, 7, 8, 8, 9, 9,10,10,11,11,12,12,12,…
0, 0, 2, 3, 4, 5, 6, 6, 7, 8, 8, 9, 9,10,10,11,11,12,12,13,13,13,14,…
0, 2, 3, 5, 6, 7, 7, 8, 9, 9,10,11,11,12,12,13,13,14,14,14,15,15,16,…
1, 3, 5, 6, 7, 8, 9,10,10,11,11,12,13,13,14,14,15,15,15,16,16,17,17,…
2, 5, 6, 8, 9,10,10,11,12,12,13,14,14,15,15,16,16,16,17,17,18,18,19,…
3, 6, 8, 9,10, 11, 12, 12, 13, 14, 14, 15, 16, 16, 17, 17, 17, 18, 18, 19, 19, 20, 20, ...
5, 7, 9,10,11,12,13,14,14,15,16,16,17,17,18,18,19,19,20,20,20,21,21,…
6, 8, 10, 11, 12, 13, 14, 15, 15, 16, 17, 17, 18, 18, 19, 19, 20, 20, 21, 21, 21, 22, 22, ...
7, 9, 11, 12, 13, 14, 15, 16, 17, 17, 18, 19, 19, 20, 20, 21, 21, 21, 22, 22, 23, 23, 23, ...
7,10,11,13,14,15,15,16,17,18,18,19,20,20,21,21,22,22,23,23,23,24,24, ...
8,10,12,13,14,15,16,17,17,18,18,19,20,20,21,21,22,22,22,23,23,24,24, ...
8,10,12,13,14,15,16,17,17,18,18,19,20,20,21,21,22,22,22,23,23,24,24,…}

  For example, “2” is shown in the 23rd column of the first row. This is because the risk level due to the brightness difference is the lowest (first stage), and the risk level due to the blinking area is the 23rd stage, the risk level is based on both the brightness difference and the blinking area. It means “2”. Also, “24” is shown in the 23rd column of the 16th row (first row). In this case as well, it means that the degree of risk on the basis of the luminance difference and the blinking area is “24”. Therefore, even if the flashing area is the same (23rd stage), it can be determined as different risk levels (2 and 24) depending on the luminance difference (1st stage and 16th stage). Similarly, when a specific example of the risk determination table 44 is considered, it can be seen that even with the same luminance difference, it can be determined as a different risk according to the blinking area.

<Result judgment / output processing>
Upon receiving the brightness classification and the blinking area calculated according to the brightness classification from the area calculation means 73, the risk determination means 75 refers to the value (8 bits: 1 to 254) of the risk determination table 44 described above. The corresponding risk value is extracted. The blink area is 14 bits for HDTV and 11 bits for SDTV. Then, the risk level judging means 75 compares the extracted numerical values, selects the maximum value among them, and outputs the numerical value as the risk level. This is referred to as a “blinking video bioeffect” (8 bits). In addition, this output shall be performed for every frame with respect to two types of input conversion values.

  Then, the risk determination means 75 uses G [i] as an array (stage (row)) counted by the luminance classification means 72 and a level (stage (column)) representing the blinking area of the area determination threshold table 43. AR [] is used to subtract the number of stages and the corresponding value PR [] is calculated. This value PR [] is the degree of risk taking the luminance difference and the blinking area as a scale. Then, the PR [] are compared with each other, and the maximum value among them is output as the blinking video body influence degree. For example, the output result RSLT is as follows.

for (i = 15; i> = 0; i--)
for (j = 31; j> = 0; j--)
if (G [i]> AR [i]) {
if (PR [i] [j]> RSLT)
RSLT = PR [i] [j];
break;
}

  Next, the configuration and processing of the image measurement circuit 7 will be described in more detail. FIG. 7 is a block diagram illustrating the configuration and processing of the image measurement circuit of FIG.

  The luminance difference calculation unit 71 includes a motion compensation unit 711 and a subtraction unit 712. The motion compensation unit 711 receives the (N + 1) th frame from the frame memory 5 (3), receives the compensation value from the motion vector FIFO storage unit 5 (4), performs motion compensation, and applies the motion compensated frame to the subtraction unit 712. It is what you pass. The subtracting means 712 subtracts the Nth frame from the frame memory 5 (3) and the motion compensated frame from the motion compensation means 711 and sends it to the comparison circuit (luminance classification means) 72.

  Then, the comparison circuit 72 compares the value from the subtracting means 711 with the value from the luminance determination threshold value table 42 for each classification according to the number of classifications in the luminance determination threshold value table 42, and belongs to which classification. Is to discriminate. Here, there are 16 levels. (1) to (16) in the comparison circuit 72 and the luminance determination threshold value table 42 correspond to the levels.

  The area calculation unit 73 includes a luminance inversion detection circuit 731, a counter 732, a constant storage unit 733, a comparison circuit 734, a switching unit 735, and counters 736 (1) to 736 (16). We have only minutes. Therefore, here, the area calculation means 73 includes 16 sets of each means.

  The luminance inversion detection circuit 731 determines whether or not luminance inversion occurs between the Nth frame and the (N + 1) th frame. The counter 732 counts the number of times that the luminance inversion detection circuit 731 has detected luminance inversion. The constant storage means 733 defines a constant “6”. Here, the constant “6” indicates the ITU-R recommendation BT. This is because, according to 1702, if the number of luminance changes exceeds 6 times, the possibility of causing a photosensitive seizure is high.

  The comparison circuit 734 compares the value of the counter 732 with the constant “6”. When the value of the counter 732 exceeds the constant “6”, the comparison circuit 734 turns on the switching unit 735 and sets the counters 736 (1) to 736 ( 16) is incremented. After this increment, the switching means 735 is turned off. In addition, the counter 732 is reset.

<Inversion edge (time axis) detection processing>
Here, the inversion edge process performed using the area calculation means 73 will be described.
The area calculation means 73 counts from 0 in both the horizontal and vertical directions of the actual image area (the part excluding the extension area) stored in the frame memory 5 (3) after the second downsampling process by the second downsampling means 63. The following processing is executed for the first even-numbered pixels (48 × 36 pixels for SDTV and 144 × 80 pixels for HDTV).

  That is, the area calculating unit 73 checks whether or not there are seven or more inversion edges (on the time axis) up to 30 frames before by the luminance inversion detection circuit 731, the counter 732, the constant storage unit 733, and the comparison circuit 734. Then, the area calculating unit 73 measures the number of pixels having the inverted edge seven times or more by the switching unit 735 and the counter 736.

  It should be noted that there are 16 edge determination thresholds (positive integers) used in the inversion edge measurement algorithm, and each threshold is measured individually. Therefore, in this case, as described above, four systems of 16 × 2 (SDTV / HDTV) × 2 (types of input conversion values) are arranged in parallel. Each threshold value is the same for both SDTV and HDTV, and is given from the same luminance determination threshold table 42.

[Inverted edge measurement algorithm]
Next, an inversion edge measurement algorithm for each pixel will be described in more detail.

  Sd0: First, the luminance inversion detection circuit 731 sets the edge flag (−1, 0, 1) of each pixel to “0” and the counter 732 of each pixel to “0”. Note that the edge flag is set in a storage unit (not shown) in the luminance inversion detection circuit 731. The edge flag (-1) and the edge flag (1) indicate that the signs are different from each other, and indicate that they are reversed by being switched. The edge flag (0) is set to an initial value.

  Sd1: Also, the luminance inversion detection circuit 731 has an output value from the subtracting means 712 that “each pixel value of the current frame” and “a pixel value compensated by a motion vector in the previous frame” are both “160” or more. If so, go to Sd7.

  Sd2: The subtracting means 712 calculates “each pixel value of the target frame” − “area value compensated with the motion vector in the previous (adjacent) frame” and outputs it to the comparison circuit 72 as “luminance difference” (−255 to 255). hand over. Then, the comparison circuit as the luminance classification unit 72 performs luminance classification with reference to the luminance determination threshold table 42, and the comparison circuit 72 according to the luminance classification operates and outputs it to the area calculation unit 73. Thereby, the luminance inversion detection unit 731 inverts the sign of “luminance” if the edge flag is “−1”. At this time, the switching means 735 is off.

  Sd3: On the other hand, the luminance inversion detection circuit 731 proceeds to Sd5 when the edge flag is other than “0” (from “−” to “1”).

  Sd4: If the “luminance difference” is equal to or greater than the threshold, the luminance inversion detection circuit 731 sets the edge flag to “−1” and the counter 732 to “1”, and proceeds to Sd7.

  Sd5: Next, the luminance inversion detection circuit 731 sets the edge flag to “1” and the counter 732 to “1” when the positive / negative inversion value of the “luminance difference” is equal to or greater than the threshold value, and proceeds to Sd6, otherwise If the luminance difference is between the positive and negative threshold values, the process proceeds to Sd7 without doing anything.

  Sd6: The luminance inversion detection circuit 731 inverts the sign of the edge flag and increments (increments) the counter 732 when the luminance difference is equal to or greater than the threshold.

Sd7: Next, if the previous frame has not reached 30 frames before the measurement start frame, the luminance inversion detection circuit 731 goes back one frame from the current frame and the previous frame, and then returns to Sd1. However, the position of the pixel in the current frame is the position of the “pixel in the previous frame” before going back.
Sd8: Then, when the previous frame reaches 30 frames before the measurement start frame, the luminance reversal detection circuit 731 determines that the comparison circuit 734 is greater than or equal to “7” where the counter 732 is larger than the constant 6 of the constant storage unit 733. In this case, the switching means 735 is turned on, the output from the comparison circuit 72 is input to the counter 736, and the counter 736 is incremented. On the other hand, when the counter is “6” or less, the next input from the comparison circuit 72 is awaited.

  On the other hand, when the luminance is counted by the counter 736 for the pixels for one frame of interest, the area classification unit 74 receives the count value, refers to the area determination threshold table 43, and determines the area for each classification. It is determined which level it belongs to, and it is determined which risk level corresponds to the classification of the risk level determination table 44.

  Further, the maximum value selection circuit (risk level determination means) 75 determines which value in the risk level determination table 44 as described above corresponds to the luminance and the area, and the maximum value among the determined numbers And the maximum value is output as the degree of risk.

  Here, the relationship between the blinking area of the screen and the blinking luminance difference will be described. FIG. 8 is a diagram for explaining the relationship between the blinking area of the screen and the blinking luminance difference. FIG. 8A shows a graph for explaining the degree of risk based on the relationship between the blinking area of the screen (vertical axis) and the blinking luminance difference (horizontal axis). As shown in this graph, even if the brightness difference is blinking, as the blinking area of the screen increases, it is considered that the degree of danger may increase as shown in risk 2, risk 4, and risk 6. (Seen prominently on the right side of the graph). Further, even if the blinking area of the screen is the same, it is considered that the degree of danger increases as the degree of danger of blinking 2, the degree of danger 4, and the degree 6 of danger as the blinking luminance difference increases.

  Here, in FIG. 8B, when each area and luminance difference reference value is one type, the danger area and safety are related to the relationship between the blinking area (vertical axis) and the blinking luminance difference (horizontal axis). A graph illustrating the region is shown as a comparative example. That is, in this case (B), the determination is made based on the threshold value of one luminance difference and the threshold value of one area. It is determined that it belongs. For this reason, in the case of (A), even if the blinking luminance difference is small and the blinking area is large, or the blinking luminance difference is large and the blinking area is small, even if the degree of danger is high, It can be determined that the degree of danger.

Next, the overall flow of the visual stimulus risk determination process of the video analysis apparatus 1 will be described. FIG. 9 shows a flowchart thereof.
The video analysis apparatus 1 converts the video signal converted into a digital signal by the A / D conversion circuit 2 into a luminance value by referring to the luminance LUT 41 by the luminance conversion circuit 3 (SA1), and the screen luminance value data Is written into the frame memory 5 for each frame.

  Next, the video analysis apparatus 1 converts the screen luminance value data written in the frame memory 5 by the first down-sampling means 62 of the image conversion circuit 6 for each (20/3) × (20/3) pixel. The average value of the block is downsampled (decimated) as each pixel value of the reduced screen (SA2).

  After that, the video analysis apparatus 1 performs block matching in the range of up / down and left / right ± 9 pixels by using the motion vector detection means 64 of the image conversion circuit 6 as one block, and searches and detects the motion vector. (SA3).

  Subsequently, the video analysis apparatus 1 uses the second downsampling unit 63 of the image conversion circuit 6 to newly calculate the average value of the four pixels of the target pixel, the right pixel, the lower pixel, and the lower right pixel as the value of each pixel. Downsampling (decimation) (SA4).

  Next, the video analysis apparatus 1 refers to the luminance determination threshold value table 42 by the image measurement circuit 7 and takes into consideration a motion vector for each luminance determination threshold value, and for each pixel up to 30 frames before. The inversion of brightness is counted (SA5). Then, the video analysis apparatus 1 determines whether or not the inversion is 7 times or more by 30 frames before the image measurement circuit 7 (SA6). If the video analysis apparatus 1 is not 7 times or more (SA6, No), the image measuring circuit 7 outputs no risk (SA7), and the process proceeds to SA10.

  On the other hand, when the number of times is seven or more (SA6, Yes), the video analyzer 1 counts the counter 736 with the determined pixel as the number of blinking pixels by the image measurement circuit 7 (SA8). Next, the video analysis apparatus 1 outputs, as the risk level, a value extracted by referring to the risk determination table 44 for the number of blinking pixels obtained for each luminance determination threshold by the image measurement circuit 7 (SA9). .

  The video analysis apparatus 1 inquires the control circuit 9 whether or not there is a next frame (SA10) every time the risk level is output (SA7, SA9). The control circuit 9 counts the number of frames input to each video analysis device 1 and notifies the video analysis device 1 to that effect when the analysis is completed for all frames. The video analysis apparatus 1 may be provided with means for executing the processing of SA10.

  When the video analysis apparatus 1 receives a notification of “next frame present” from the control circuit 9 (SA10, Yes), it returns the process to SA1, and receives a notification that is not “next frame present” (SA10, No). ), The process is terminated.

  In the above embodiment, the case of HDTV has been described as an example. However, the same processing can be performed for SDTV. FIG. 10 shows a block diagram of a configuration of a luminance conversion circuit corresponding to FIG. As shown in FIG. 10, the SDTV has the same configuration as the HDTV (FIG. 3). The only difference is that the luminance LUT 41 is different. Also, the image conversion circuit 6 and the image measurement circuit 7 have the same configuration as in the case of HDTV, and thus description thereof is omitted. That is, SDTV and HDTV differ only in the numerical values of the luminance LUT 41, the luminance determination threshold table 42, the area determination threshold table 43, and the risk determination table 44.

  Finally, a configuration for parallel processing in the case of HDTV and SDTV will be described. FIG. 11 shows a block diagram of a video analysis system configured to perform the visual stimulus risk determination process in parallel. The video analysis system 100 includes video analysis devices 1A and 1B and a control circuit 8A in a housing (not shown), and an HDTV video signal input terminal 9A and an SDTV video signal input terminal 9B as viewed from the housing. And.

  The video analysis apparatus 1A processes an HDTV video signal. The video analysis apparatus 1B processes SDTV video signals. The control circuit (control means) 8A controls the operations of the video analysis apparatuses 1A and 1B. The HDTV video signal input terminal 9A inputs an HDTV video signal (video signal) to the video analyzer 1A. The SDTV video signal input terminal 9 inputs an SDTV video signal to the video analyzer 1B.

  The video analysis system 100 having the above configuration processes in parallel the analysis of the SDTV video signal by the video analysis device 1B and the analysis of the HDTV video signal by the video analysis device 1A, under the control of the control circuit 9A. Further, since each of the video analysis apparatuses 1A and 1B independently processes the analysis of each video signal, the gamma characteristics of both video signals can be analyzed simultaneously. Therefore, according to the video analysis system 100, it is possible to efficiently analyze the visual stimulus risk level.

  In the above-described embodiment, it has been described that each determination criterion is given as a table, but it may be given by the following calculation formula or by a program. Here, the possibility ρ of causing a seizure can be evaluated by, for example, the following formula, assuming that the blinking luminance difference x is the area ratio y. The risk determination table 44 described in the above embodiment can also be calculated according to the following formula.

ρ (x, y)
= (− 3 × 10 ¹⁶ · x ³ + 3 × 10 ⁻⁴ · x ² + 7 × 10 ⁻⁴ · x−3 × 10 ⁻⁴ )
× (−0.7241y ² + 1.8124y + 0.0878)

  Each of the circuits (means) is not only configured as an electronic circuit, particularly an IC chip, but also configured as a visual stimulus risk determination program and a computer that executes the visual stimulus risk determination program. Also good. Therefore, the video analysis apparatus 1 can be realized by causing a general computer to execute a program and operating an arithmetic device or a storage device in the computer. The visual stimulus risk determination program can be distributed via a communication line, or can be written and distributed on various recording media.

  Therefore, according to the video analysis apparatus of the embodiment, it is possible to determine the degree of danger of visual stimulation by a relatively safe video or a very dangerous video. For this reason, the video analyzer registers in advance in the luminance determination threshold table 42, the risk determination table 43, and the risk determination table 44 characteristic data about visual stimuli such as blinking light, red light, and a graphic with strong contrast. Therefore, it is possible to easily and accurately determine the risk associated with them.

It is a block diagram explaining the structure of the video-analysis apparatus of embodiment which concerns on this invention. FIG. 2 is a block diagram illustrating a more detailed configuration of the luminance change circuit of FIG. 1. It is a flowchart of the flow of a process of the brightness | luminance change circuit shown in FIG. FIG. 2 is a block diagram of the image conversion circuit in FIG. 1. 6 is a flowchart of a process flow of the image conversion circuit in FIG. 4. It is a block diagram of a structure of the image measurement circuit of FIG. It is a block diagram explaining the structure and process of the image measurement circuit of FIG. It is a figure explaining the relationship between the blink area of a screen, and a blink brightness difference. It is a flowchart explaining the whole flow of the visual stimulation risk degree determination process of a video analyzer. FIG. 4 is a block diagram of a configuration of a luminance conversion circuit (SDTV) corresponding to FIG. 3. It is a block diagram of the image | video analysis system of the structure which processes a visual stimulation risk degree determination process in parallel.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Video analyzer 3 Luminance conversion circuit 4 Flash memory 5 Frame memory 6 Image conversion circuit 7 Image measurement circuit 8 Judgment reference change means 9 Control circuit 41 Luminance LUT
42 Luminance determination threshold table (luminance classification storage means)
43 Area determination threshold table (area classification storage means)
44 Risk level determination table (risk level criterion storage means)
71 brightness difference calculating means 72 brightness classifying means 73 area calculating means 74 area classifying means 75 risk degree judging means

Claims (5)

A video analysis device that analyzes a video to be displayed on a display and determines a risk level indicating a possibility of causing a photosensitive seizure to a viewer by a visual stimulus by the video,
Luminance classification storage means for storing luminance classification as a criterion for classifying pixels into a plurality of sets according to the luminance difference between the pixels;
Area classification storage means for storing area classification as a criterion for classifying pixels into a plurality of sets according to the size of the blinking area;
A risk criterion storage means for storing a criterion for determining a risk indicating a possibility of causing a photosensitive seizure according to the luminance classification and the blinking area;
A luminance difference calculating means for calculating a luminance difference between arbitrary pixels located at the same coordinates between an arbitrary target frame in a frame constituting the video and its adjacent frame;
Luminance classification means for classifying pixels of the luminance difference calculated by the luminance difference calculation means according to the luminance classification stored in the luminance classification storage means;
An area calculation unit that counts the number of pixels for each luminance classification classified by the luminance classification unit, and calculates the counted value as a blinking area according to the luminance difference;
Area classification means for classifying the blinking area calculated by the area calculation means according to the area classification stored in the area classification storage means;
For each of the brightness classifications, the risk level determination means is referred to according to the classification results of the brightness classification means and the area classification means, and the risk level is determined in multiple stages according to the determination criteria for determining the risk level. A risk determination means for selecting and outputting the highest risk level among the risk levels determined in multiple stages for each of the luminance classifications ;
A video analysis apparatus comprising: Comprising a downsampling means for thinning and sampling pixels of the video ,
2. The video analysis apparatus according to claim 1, wherein pixels downsampled by the downsampling unit are used as pixels in the luminance difference calculation unit, the luminance classification unit, and the area calculation unit . Luminance conversion means for converting the digitized video signal of the video into a luminance value ;
The video analysis apparatus according to claim 1, wherein the luminance difference calculation unit calculates a difference in luminance value between the pixels after luminance conversion by the luminance conversion unit. In order to analyze the video to be displayed on the display and determine the degree of risk that the visual stimulus from the video may cause the viewer to cause a light-sensitive seizure,
A luminance difference calculating means for calculating a luminance difference between arbitrary pixels located at the same coordinates between an arbitrary target frame in a frame constituting the video and its adjacent frame;
A luminance classification is read from a luminance classification storage unit that stores luminance classifications as determination criteria for classifying pixels into a plurality of sets according to the luminance difference between pixels, and is calculated by the luminance difference calculation unit according to the luminance classification Luminance classification means for classifying the pixels of the luminance difference obtained,
An area calculating means for counting the number of pixels for each luminance classification classified by the luminance classification means, and calculating the counted value as a blinking area according to the luminance difference;
An area classification is read from an area classification storage unit that stores an area classification as a criterion for classifying pixels into a plurality of sets according to the size of the blinking area, and is calculated by the area calculation unit according to the area classification. Area classification means for classifying the blinking area,
For each luminance classification, according to the classification result of the luminance classification means and the area classification means, a criterion for determining a risk level indicating the possibility of causing a light-sensitive seizure according to the luminance classification and the blinking area is stored. Referring to the risk determination criterion storage means to determine the risk in multiple stages according to the determination criteria for determining the risk, the highest among the risk determined in multiple stages for each luminance classification Risk level judging means for selecting and outputting the risk level,
Visual stimulus risk determination program, characterized by functioning as A video analysis system for processing an SDTV video signal and an HDTV video signal in parallel by the video analysis device according to any one of claims 1 to 3 ,
A first video analysis device for analyzing the SDTV video signal;
A second video analysis device for analyzing the HDTV video signal;
Control means for processing in parallel the analysis of the SDTV video signal by the first video analysis device and the analysis of the HDTV video signal by the second video analysis device;
A video analysis system characterized by comprising:

Images