US20050155054A1 - Summarization of sumo video content - Google Patents

Abstract

Summarization of video content including sumo.

Description

BACKGROUND OF THE INVENTION
• The present invention relates to summarization of video content including sumo.
• The amount of video content is expanding at an ever increasing rate, some of which includes sporting events. Simultaneously, the available time for viewers to consume or otherwise view all of the desirable video content is decreasing. With the increased amount of video content coupled with the decreasing time available to view the video content, it b1ecomes increasingly problematic for viewers to view all of the potentially desirable content in its entirety. Accordingly, viewers are increasingly selective regarding the video content that they select to view. To accommodate viewer demands, techniques have been developed to provide a summarization of the video representative in some manner of the entire video. Video summarization likewise facilitates additional features including browsing, filtering, indexing, retrieval, etc. The typical purpose for creating a video summarization is to obtain a compact representation of the original video for subsequent viewing.
• There are two major approaches to video summarization. The first approach for video summarization is key frame detection. Key frame detection includes mechanisms that process low level characteristics of the video, such as its color distribution, to determine those particular isolated frames that are most representative of particular portions of the video. For example, a key frame summarization of a video may contain only a few isolated key frames which potentially highlight the most important events in the video. Thus some limited information about the video can be inferred from the selection of key frames. Key frame techniques are especially suitable for indexing video content but are not especially suitable for summarizing sporting content.
• The second approach for video summarization is directed at detecting events that are important for the particular video content. Such techniques normally include a definition and model of anticipated events of particular importance for a particular type of content. The video summarization may consist of many video segments, each of which is a continuous portion in the original video, allowing some detailed information from the video to be viewed by the user in a time effective manner. Such techniques are especially suitable for the efficient consumption of the content of a video by browsing only its summary. Such approaches facilitate what is sometimes referred to as “semantic summaries”.
• What is desired, therefore, is a video summarization technique suitable for video content that includes sumo.
BRIEF DESCRIPTION OF THE DRAWINGS
• FIG. 1 is an exemplary flowchart for play detection.
• FIG. 2 is an exemplary illustration of a pre-bout scene in sumo.
• FIG. 3 is a technique for detecting a start frame of a sumo “play.”.
• FIG. 4 is a pre-bout scene in sumo.
• FIG. 5 illustrates the skin color and ring color ofFIG. 4.
• FIG. 6 illustrates binarized skin color ofFIG. 5.
• FIG. 7 is a horizontal projection ofFIG. 6.
• FIG. 8 is a vertical projection ofFIG. 6.
• FIGS. 9A-9C is a series of sequential images in a video clip showing two sumo contestants colliding.
• FIG. 10 is an illustration of temporal evidence accumulation.
• FIG. 11 is an illustration of color histogram differences.
• FIG. 12 is an illustration of absolute pixel-to-pixel differences in luminance domain.
• FIG. 13 illustrates scene cut detection.
• FIG. 14 illustrates names in a sumo video.
• FIGS. 15A-15C illustrate audio segments of different plays.
• FIG. 16 illustrates forming a multi-layered summary of the original video sequence.
• FIG. 17 illustrates the video summarization module as part of a media browser and/or a service application.
• FIG. 18 illustrates a video processing system.
• FIG. 19 illustrates an exemplary overall structure of the sumo summarization system.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
• Sumo, the national sport of Japan, is tremendously popular in eastern Asia and is growing in popularity elsewhere in the world. Sumo is a sport comprising bouts in which two contestants meet in a circular ring 4.55 meters in diameter. The rules of Sumo are uncomplicated. After the contestants and a referee have entered the circular ring, the bout begins with an initial charge—called a “tachiai”—where each contestant rushes towards, then collides with, the other. The bout will end when one of the contestant loses by either stepping outside the circular ring or touching the ground with any part of the contestant's body other than the soles of the feet. Aside from a limited number of illegal moves, such as gouging the opponent's eyes, striking with a closed fist, or intentionally pulling at the opponent's hair, there are no rules that govern a sumo bout.
• Sumo participants may compete against each another in one of a number of tournaments. Japan sponsors six sanctioned Grand Sumo tournaments, held in odd-numbered months throughout the year, in which competitive sumo contestants face one another with the opportunity for advancement in rank. Sumo contestants are ranked under a strict meritocracy; winning bouts in these sanctioned tournaments improves a competitor's rank while losing bouts diminishes that rank. Aside from the six sanctioned tournaments, a number of exhibition tournaments—called Jungyo—are scheduled throughout the year.
• Though a sumo tournament will typically take place over several weeks with bouts scheduled throughout each day, most bouts of interest, i.e. those involving higher ranked contestants, are scheduled to begin late afternoon when live television broadcasts of the tournament occur. These portions of the sumo tournaments usually last 2-3 hours each day and are often video recorded for later distribution or for re-broadcast.
• Though such a video of a sumo tournament might typically last about 2-3 hours, only about ten minutes turns out to include time during which two players are in a bout. An individual sumo bout is brief; the typical bout will end with the initial collision, though a rare bout might last two to three minutes. Interspersed between bouts are a large number of ceremonies that precede and follow each bout.
• Though brief, the time intervals during which a bout is proceeding are intense and can captivate those in the viewing audience, many of whom are able to identify a myriad of named sumo techniques that may occur in rapid succession. Such techniques include a “kekaeshi” (a foot-sweep), a “kubinage” (a head-lock throw), and an “izori” (a technique where a contestant crouches below the opponent's rush, grabbing one of the opponent's legs, lifting the opponent upon the shoulders and falling backwards), as well as some sixty five to seventy more named sumo techniques or occurrences.
• The remaining time during the sumo tournament is typically not exciting to watch on video. Such time would include for example inter-bout changes of players, pre-bout exercises and ceremonies, post-bout ceremonies and in the case of broadcast, nearly endless commercials. While it may indeed be entertaining to sit in an arena for several hours for a sumo tournament, many people who watch a video of a sumo tournament find it difficult to watch all of the tournament, even if they are rabid fans. Further, the tournaments are held during daytime hours, hence many fans are unable to attend a tournament or to watch a live broadcast due to work. Such fans may nonetheless be interested in watching specific bouts or some other condensed version of the tournament. Thus a video summarization of the sumo tournament that provides a summary of the tournament having a duration shorter than the original sumo video, may be appealing to many people. The video summarization should provide nearly the same level of the excitement (e.g. interest) that the original game provided.
• Upon initial consideration, sumo would not be a suitable candidate to attempt automated video summarization. Initially, there are nearly an endless number of potential moves that may occur that would need to be accounted for in some manner. In addition, each of these moves may involve significant player motion that is difficult to anticipate, difficult to track, and is not consistent between plays. In addition, the players are flesh toned and the ring is likewise generally flesh toned making identification of the events difficult. Based upon such considerations it has been previously considered impractical, if not impossible, to attempt to summarize sumo.
• It is conceivably possible to develop highly sophisticated models of a typical sumo video to identify potentially relevant portions of the video. However, such highly sophisticated models are difficult to create and are not normally robust. Further, the likelihood that a majority of the highly relevant portions of the sumo video will be included in such a video summarization is low because of the selectivity of the model. Thus the resulting video summarization of the sumo tournament may simply be unsatisfactory to the average viewer.
• After consideration of the difficulty of developing highly sophisticated models of a sumo video to analyze the content of the sumo video, as the sole basis upon which to create a sumo summarization, the present inventors determined that this technique is ultimately flawed as the models will likely never be sufficiently robust to detect all the desirable content. Moreover, the number of different types of model sequences of potentially desirable content is difficult to quantify. In contrast to attempting to detect particular model sequences, the present inventors determined that the desirable segments of the sumo match are preferably selected based upon a “play”. A “play” may be defined as a sequence of events defined by the rules of sumo. In particular, and in one aspect, the sequence of events of a “play” may generally include the time between which the players line up to charge one another and one player loses the bout by either stepping outside the sumo ring or touching the clay surface with a part of his body other than the soles of the feet. A play may also selectively include certain pre-bout ceremonies or events, such as the time during which the contestants throw salt in the ring or stare at one another prior to charging. Normally the “play” should include a related series of activities that could potentially result in a victory by one contestant and a loss by the other contestant.
• It is to be understood that the temporal bounds of a particular type of “play” does not necessarily start or end at a particular instance, but rather at a time generally coincident with the start and end of the play or otherwise based upon, at least in part, a time (e.g., event) based upon a play. For example, a “play” starting with the contestants throwing salt into the ring may include the times during which the contestants charge each other. A summarization of the video is created by including a plurality of video segments, where the summarization includes fewer frames than the original video from which the summarization was created. A summarization that includes a plurality of the plays of the sumo match provides the viewer with a shorted video sequence while permitting the viewer to still enjoy the game because most of the exciting portions of the video are provided, preferably in the same temporally sequential manner as in the original sumo video.
• Referring toFIG. 1, a procedure for summarization of a sumo video includes receiving avideo sequence20 that includes at least a portion of a sumo match.Block22 detects the start of a play of a video segment of a plurality of frames of the video. After detecting the start of the play, block24 detects the end of the play, thereby defining a segment of video between the start of the play and the end of the play, namely, a “play”.Block26 then checks to see if the end of the video (or the portion to be processed) has been reached. If the end of the video has not been reachedblock26 branches to block22 to detect the next play. Alternatively, if the end of the video has been reached then block26 branches to thesummary description28. The summary description defines those portions of thevideo sequence20 that contain the relevant segments for the video summarization. The summary description may be compliant with the MPEG-7 Summary Description Scheme or TV-Anytime Segmentation Description Scheme. A compliant media browser, such as shown inFIG. 17, may apply the summary description to the input video to provide summarized viewing of the input video without modifying it. Alternatively, the summary description may be used to edit the input video and create a separate video sequence. The summarized video sequence may comprise the selected segments which excludes at least a portion of the original video other than the plurality of segments. Preferably, the summarized video sequence excludes all portions of the original video other than the plurality of segments.
• FIG. 1 is intended to show a basic procedure for obtaining such a summary, where the summary description contains only the start and end points of the detected plays. The summarization shown inFIG. 1 is primarily a low-level one, though in more complex situations it may contain other information, i.e. names of contestants etc. The benefit of a low-level summary is that it provides sufficient detail for people to appreciate a game from the summary. The low-level summary may then form the basis for a higher level summarization, if desired. As one example, a higher level summary can be obtained by keeping only those plays receiving loud audience acclaims, achieved by adding an audio analysis procedure. Alternatively, in combination with a caption detection/recognition module, a summary can be obtained of only those plays containing a specific contestant. A yet higher summary level may contain only key frames from the plays for indexing purposes.
• One component of the summarization procedure depicted inFIG. 1 is the detection of an event, or “play.” If the start and end points of all plays are detected, then the system may string all the plays together to obtain a summary from the original video and perform some post processing to smooth the transition boundaries, such as using dissolving techniques to reduce abrupt change between plays and smoothing the audio filed for better auditory effects. Further, the summary should ideally contain only those segments comprising a “play” as earlier defined, thus providing a compact representation of the original tournament; the user can spend only a few minutes to watch it, yet almost all of the excitement of the original game can be appreciated.
• One of the difficulties in the detection of a “play” in a sumo broadcast is that frames in one play may sweep a large range of color, yet all the frames belong to the same event, and form an uninterrupted video clip. Thus a generic summarization scheme that uses, for example, a color histogram as the cue for key frame detection or scene classification, may not be particularly effective. In light of such difficulties, the present inventors have developed an alternate method for detecting a “play” that is specifically tailored to sumo content.
• Still referring toFIG. 1, a summary is to be obtained by first detecting the boundaries of a “play.” In a sumo bout, two contestants meet in a ring 4.55 meters across. Though they wear silk belts around their waists, the players are otherwise unclothed. There are strict rules as to where the players and the referee, called a “Gyoji,” are to stand in the moments immediately proceeding the initiation of the bout. Cameras are situated at fixed locations around the ring capture the sumo bout. The cameras can typically pan, tilt, and zoom. The primary camera typically is situated opposite to the side where the referee stands. Thus a bout usually starts with a scene as illustrated inFIG. 2, and the bout will almost always be broadcast in its entirety by the primary camera from this vantage. Video captured by any other camera is typically used exclusively for replays, player close-ups, or post-bout ceremonies, all of which take place after the bout has ended. This format is adhered to because the primary camera can best cover the action of the bout, which usually lasts for mere moments, making it impractical to switch camera angles during a bout.
• Based on these observations, the inventors have developed a model for “play” detection. A play starts with a scene as inFIG. 2. The time between the scene cut at the end of a current play and the start of the following play is not usually exciting and can thus be excluded from a compact summary. Note that a scene like that shown inFIG. 2 is typically merely a necessary condition, not a sufficient condition. In a sumo tournament, there are many pre-game ceremonies that result in a scene like that shown inFIG. 2, but the contestants, are not yet ready to initiate the bout. Thus in order to detect the start of a play, in addition to finding a scene like that depicted inFIG. 2, it should determine whether the scene is an immediate precursor to the start of a bout. One test would be to determine whether the contestants charge one another and collide, because that is how each bout begins. In other words, the methodology of detecting whether the start of a “play” has occurred involves locating a frame similar to that shown inFIG. 2 then applying a test to determine whether the frame immediately precedes the start of a bout.
• The location of frames similar to that shown inFIG. 2 may be based on the anticipated characteristics of the image, as opposed to an actual analysis of the events depicted in the video. Under the assumption that a camera gives a typical start frame like that shown inFIG. 2, one can observe that the lower part of such frame contains the stage in which the sumo ring is defined. The stage in the lower part of the frame is usually of fixed color and lighter than the generally dark color of the upper part of the frame. This is usually true because a sumo stage is to be constructed according to the same specifications. Further, in a sumo tournament the lights are usually focused on the stage which give tends to shroud the background in darkness. In addition, each bout is preceded with the two contestants facing one another in a symmetric position about the center of the ring with the referee to the side and between the contestants and the primary camera facing the referee.
• The color of the stage can be estimated from sample data; given a set of sample frames containing the stage, a set of parameters can give an estimate for the stage color. Detecting the players is a more difficult task. Theoretically, one could use complex methods such as those explicitly modeling the shape of a human body. To achieve fast computation, the present inventors have identified a simpler method in describing a player: a player is represented by a color blob with skin tone. Thus assuming that an estimate for skin tone is obtained, two blobs corresponding to the two respective players could be segmented. As mentioned earlier, in a Sumo broadcast, there are pregame ceremonies that could result in frames like a start frame. To enable this type of false alarm to be eliminated, the players should be tracked after they are detected to see if they move towards each other and eventually collide with each other, as would occur at the beginning of a “play” as earlier defined.
• One method for detecting the beginning of a play may proceed as shown inFIG. 3. Given a stage color description Cs and skin tone description Ck, a video frame image IM can be examined to determine whether the image represents the beginning of a “play.” The color descriptions, may be for example, a single color, a range of colors, a set of colors, in one or more color spaces. First, the image is examined to determine if it has a dark upper portion and a lower portion dominated (25% or more, 50% or more, or 75% or more) by the color Cs+Ck. If not, then the image is determined as a non-start frame. If yes, then the image is examined to determine whether there are two dominant (25% or more, 50% or more, or 75% or more) color blobs of color Ck, nearly symmetric to each other with respect to a generally center column (+/−20% of the width of the frame off center) of the frame. If not, then the image is determined as a non-start frame. If yes, subsequent frames are examined to determine whether the two dominant color blobs move towards, and eventually collide with, one another. If so, the original frame image IM is determined a start frame, otherwise it is determined not to be a start frame. The technique may be modified to include fewer tests or additional tests, in the same or a different sequence.
• It turns out that a difficult part of this method is to segment the player blobs from the stage because the stage color Cs and the skin tone Ck are overlapping in typical color space. It is impossible to perfectly separate skin from the stage only using color information, which means that the player detection is always imperfect and the players are usually detected as fragmented pieces. In fact, this is inevitable, considering that the players often wear belts of various non-skin tone color. If a single blob is to be detected for each player, then an additional module must be used to group the fragmented pieces. This module may again introduce additional inaccuracies, aside from the demand for additional computation.
• To avoid the computational burden and potential inaccuracies of such a grouping procedure, the present inventors discovered that the foregoing method for detecting the beginning of a play may be implemented by representing and tracking the blobs through their one-dimensional projections.FIG. 4 shows a candidate image IM that is a representative start frame of a sumo “play” as earlier defined, and thus should be detected by the summarization procedure shown inFIG. 1. Given the stage color description Cs and the skin tone description Ck, the candidate image shown inFIG. 4 may be reduced to the image shown inFIG. 5 where white pixels indicate a place where there is a pixel in the candidate image corresponding to either the stage color Ck or the skin color Cs. The black pixels represent the dark background areas of the candidate image. The image may be further decomposed using skin-tone based segmentation to isolate those portions of the image corresponding to the skin color Cs. A binary image, shown inFIG. 6 may be used to represent the obtained body parts, in which numeral ones represent a pixel of that location representing skin in the original image. This binary image may be projected along vertical and horizontal axes, shown inFIGS. 7 and 8, respectively. The analysis of the blob may be performed on those projections. The proposed projection behaves effectively like an integration process, which makes the algorithm less sensitive to imperfection in the skin/stage segmentation. Note that in these projections, small and isolated peaks have been suppressed.
• Ideally, a real start frame will result in two peaks of similar size in the vertical projection, nearly symmetric about the center column of the image, as shown inFIGS. 7 and 8, the horizontal projection of the binary image, may be used to check whether the two blobs are symmetric about a center column of the image. In subsequent frames, these two peaks should move closer and closer, eventually converging, as illustrated byFIGS. 10A, 10B, and10C.
• The foregoing method relies mainly on color cues, and prior knowledge about the stage color Cs and the skin tone Ck are assumed. However, it is also possible to calibrate the colors for a specific bout or tournament. With other inputs such as a human operator's interactions, the calibration is of course easy to do. Without any human interaction, statistical models can still be used to calibrate the color. If a series of start scene candidates has been obtained, statistical outliers in this set can be detected with prior coarse knowledge about Cs and Ck. The remaining candidate frames can then be used to estimate the specifics of the colors. With the colors calibrated, the start-of-play detection can be performed more accurately.
• The foregoing method is able to detect start frames successfully in most situations. However, if the detection of a start frame is declared after finding only one candidate frame, then the method may be susceptible to false-positives. By examining a set of consecutive frames (or other temporally related frames) and accumulating evidence, the system can reduce the false-positive rate. Referring toFIG. 10, the following approach may be used to achieve temporal evidence of accumulation: when detecting the start of a “play”, a sliding window of width w is used (e.g., w frames are considered at the same time). A start is declared only if more than p out of the w frames in the current window are determined to be start scene candidates, as previously described. A suitable value of p is such that p/w=70%. Other statistical measures may be used of a fixed number of frames or dynamic number of frames to more accurately determine start scenes.
• While the start of a “play” may be found according to the aforementioned method, the end of a “play” can occur in a variety of different ways due to the numerous techniques used to either force the opposing contestant to the ground or out of the ring. Image analysis techniques may be used to analyze the image content of the frames after the beginning of a bout to attempt to determine what occurred, but with the nearly endless possibilities and the difficulty of interpreting the content of the frames, this technique is at least, extremely difficult and computationally intensive. In contrast to attempting to analyze the content of the subsequent frames of a potential play, the present inventors determined that a more efficient manner for the determination of the extent of a play in sumo is to base the end of the play on camera activities. After analysis of a sumo video the present inventors were surprised to determine that the approximate end of a play may be modeled by scene changes, normally as a result of switching to a different camera or a different camera angle. The different camera or different camera angle may be modeled by determining the amount of change between the current frame (or set of frames) to the next frame (or set of frames).
• Referring toFIG. 11, a model of the amount of change between frames using a color histogram difference technique for an exemplary 1,000 frame video sumo clip is shown. The peaks typically correspond to scene cuts. Unfortunately,FIG. 11 demonstrates, some scene cuts, like the one depicted at around frame325, the camera break produces a relatively low peak in the color histogram difference curve, causing potential failure in scene cut detection.
• To solve this problem, the inventors have discovered that the use of color histogram differences in conjunction with the sum of absolute pixel-to-pixel differences in the luminance domain is more effective when detecting a scene cut. To gain robustness in using the sum of absolute pixel-to-pixel differences, the luminance images are first down-sampled, or smoothed.FIG. 13 shows the sum of absolute pixel-to-pixel luminance differences for the same video clip as shown inFIG. 11.
• Even with the aforementioned technique there may be some false detections which do not correspond to a real play. Also, there are situations in which a play is broken into two segments due to for example, dramatic lighting fluctuations (mistaken by the system as a scene cut). Some of these problems can be remedied by post-processing. One example of a suitable post processing technique is if two plays are only separated by a sufficiently short time duration, such as less than a predetermined time period, then they should be connected as a single play. The time period between the two detected plays may be included within the total play, if desired. Even if the two detected plays are separated by a short time period and the system puts the two plays together, and they are in fact two separate plays, this results in an acceptable segment (or two plays) because it avoids frequent audio and visual disruptions in the summary, which may be objectionable to some viewers. Another example of a suitable post processing technique is that if a play has a sufficiently short duration, such as less than 2 seconds, then the system should remove it from being a play because it is likely a false positive. Also, post-processing may be applied to smoothen the connection between adjacent plays, for both video and audio.
• Sumo video may also include gradual transitions between plays and other activities, such as commentary. These gradual transitions tend to be computationally complex to detect in the general case. However, in the case of sumo it has been determined that detecting gradual transitions based upon the color histogram differences is especially suitable. Other techniques may likewise be used. Referring toFIG. 13, the preferred technique may include starting from a start-of-play time (t_o) and looking forward until a sufficiently large scene change is detected or until time t_o+t_pis reached, whichever occurs first. T_prelates to the maximum anticipated play duration and therefore automatically sets a maximum duration to the play. This time period for processing to locate gradual transitions is denoted as t_clean—cut. If t_clean—cut<t_lowthen the system will not look for a gradual scene cut and set the previously detected scene cut as the end of the play. This corresponds to an anticipated minimum time duration for a play and t_lowis used to denote the minimum time period. Otherwise, the system looks for the highest color histogram difference in the region t_low, t_clean—cutor other measure of a potential scene change. This region of the segment is from the minimum time duration to the next previously identified scene cut. This identifies the highest color histogram difference in the time duration which may be a potential scene change. The time of the highest color histogram difference is identified at t₁. In a neighborhood of t₁, [t₁−c₁, t₂+c₂], a statistical computation is performed, such as computing the mean m₁and the standard deviation F of the color histogram differences. C₁and c₂are constants or statistically calculated temporal values for the region to examine around the highest color histogram difference. A mean filtering emphasizes regions having a relatively large difference in a relatively short time interval. If the color histogram differences at t₁exceeds m₁+c₃*F₁, where c₃is a constant (or otherwise) and some of its neighbors (or otherwise) are sufficiently large, then the system considers a gradual transition to have occurred at around time (frame) t₁. The play is set to the shorter of the previously identified scene cut or the gradual transition, if any.
• The summary obtained by the method described above contains only play segments from the original video. Even though a Sumo fan may be able to quickly recognize the players after they appear, it may help a viewer to follow the game better if we detect those pre-play frames that contains player's names. An example of such type of frames is given inFIG. 14.
• There are various ways of detecting overlaid graphical text content from an original image or video. In this application, the problem is one of detecting Kanji (Chinese characters used in Japanese) in images. With sufficient sample data, the system may train a convolution neural network to perform this task. In Sumo broadcasting there are a few special patterns that are typically adopted in presenting the graphical characters. For example, the names of the two players are the biggest characters. Also, it appears that the names normally appear in white (or substantial contrast to the background). This is probably due to the fact that the names are usually overlaid on a dark scene of the sumo stadium. In addition the graphical information is symmetric with respect to the center column, with one player's information on the left, and the other player's information on the right. The characters read vertically from top to bottom.
• These special patterns can be utilized to facilitate a neural network based character detection module. The system may include an algorithm to find frames with these patterns. The present inventors have found that the following set of rules may successfully detect frames with the desired player names in a video: (1) the frame has white blocks that are nearly symmetrically distributed about the center column of the image; (2) except for these white blocks, there should be no other white areas of significant size in the frame; (3) these white blocks persist for at least a few seconds; and (4) the set of frames with persistent white blocks proceeds to the start of a play. One or more of these rules may be included, as desired.
• After the frames with the player names are detected, the system may add them to their respective plays and obtain a new summary. Unlike the baseline summary obtained before, in this new summary, there are a few seconds of video like that inFIG. 14 for introducing each play. Thus the new summary is easier to follow.
• If desired, a slow motion replay detection module may be incorporated. The system detects if a slow motion replay has occurred, which normally relates to important events. The system will capture the replays of plays, the same as the typical non-slow motion replay (full speed), if the same type of camera angles are used. The play segments detected may be identified with multiple characteristics, namely, slow motion replay-only segments, play only segments without slow motion replay segments, and slow motion replay that include associated full speed segments. The resulting summary may include one or more of the different selections of the aforementioned options, as desired. For example, the resulting summary may have the slow-motion replays removed. These options may likewise be user selectable.
• While an effective summarization of a sumo video may be based on the concept of the “play”, sometimes the viewer may prefer an even shorter summarization with the most exciting plays included. One potential technique for the estimation of the excitement of a play is to perform statistical analysis on the segments to determine which durations are most likely to have the highest excitement. However, this technique will likely not provide sufficiently accurate results. Further, excitement tends to be a subjective measure that is hard to quantify. After further consideration the present inventors came to the realization that the audio provided together with the video provides a good indication of the excitement of the plays. For example, the volume of the response of the audience and/or the commentators provides a good indication of the excitement. The louder audience and/or commentator acclamations, the greater the degree of excitement.
• Referring toFIGS. 15A-15C, an exemplary illustration is shown of audio signals having a relatively quiet response (FIG. 15A), having a strong response (FIG. 15B), and having an extremely strong response (FIG. 15C). In general, it has been determined that more exciting plays have the following audio features. First, the mean audio volume of the play is large. The mean audio volume may be computed by defining the mean volume of a play as$v=\frac{1}{N}\sum _{i=0}^{N-1}{\mathrm{<figure-callout\; label="S"\; filenames="US20050155054A1-20050714-D00004.png,US20050155054A1-20050714-D00009.png"\; state="\{\{state\}\}">S</figure-callout>}}^2\left(i\right)$
  where S(i) is the i-th sample, and the N is the total number of samples in the play. Second, the play contains more audio samples that have middle-ranged magnitudes. The second feature may be reflected by the percentage of the middle-range-magnituded samples in the play, which may be computed as$P=\frac{1}{N}\sum _{i=0}^{N-1}I\left(\mathrm{\_s}\left(i\right)\_>\mathrm{t1}\mathrm{and\_s}\left(i\right)\_<\mathrm{t2}\right)$
  with I( ) being the indicator function (I(true)=1, and I(false)=0), t1 and t2 are two thresholds defining the middle range.
• Referring toFIG. 16, the first layer of the summary is constructed using the play detection technique. The second and third layers (and other) are extracted as being of increasingly greater excitement, based at least in part, on the audio levels of the respective audio of the video segments. Also, it would be noted that the preferred audio technique only uses the temporal domain, which results in a computationally efficient technique. In addition, the level of the audio may be used as a basis for the modification of the duration of a particular play segment. For example, if a particular play segment has a high audio level then the boundaries of the play segment may be extended. This permits a greater emphasis to be placed on those segments more likely to be exciting. For example, if a particular play segment has a low audio level then the boundaries of the play segment may be contracted. This permits a reduced emphasis to be placed on those segments less likely to be exciting. It is to be understood that the layered summarization may be based upon other factors, as desired.
• Referring toFIG. 17, the video summarization may be included as part of an MPEG-7 based browser/filter, where summarization is included within the standard. The media summarizer may be as shown inFIG. 1. With different levels of summarization built on top of the aforementioned video summarization technique, the system can provide the user with varying levels of summaries according to their demands. Once the summary information is described as an MPEG-7 compliant XML document, one can utilize all the offerings of MPEG-7, such as personalization, where different levels of summaries can be offered to the user on the basis of user's preferences described in an MPEG-7 compliant way. Descriptions of user preferences in MPEG-7 include preference elements pertaining to different summary modes and detail levels.
• In the case that the summarization is performed at a server or service provider, the user downloads and receives the summary description encoded in MPEG-7 format. Alternatively, in an interactive video on demand (VOD) application, the media and its summary description reside at the provider's VOD server and the user (e.g., remote) consumes the summary via a user-side browser interface. In this case, the summary may be enriched further by additional information that may be added by the service provider. Further, summarization may also be performed by the client.
• Referring toFIG. 18, the output of the module that automatically detects important segments may be a set of indices of segments containing plays and important parts of the input video program. A description document, such as an MPEG-7 or TV-Anytime compliant description is generated in The Description Generation module. Summary segments are made available to the Post-Processing module by The Extraction of Summary Segments module which processes the input video program according to the description. A post-processing module processes the summary Segments and/or the description to generate the final summary video and final description. The post-processing module puts the post-processed segments together to form the final summary video. The post-processing module may transcode the resulting video to a format different that of the input video to meet the requirements of the storage/transmission channel. The final description may also be encoded, e.g., binarized if it is generated originally in textual format such as XML. Post-processing may include adding to the original audio track a commentary, insertion of advertisement segments, or metadata. In contrast to play detection, post-processing may be completely, or in part, manual processing. It may include, for example, automatic ranking and subset selection of events on the basis of automatic detection of features in the audio track associated with video segments. This processing may be performed at the server and then the resulting video transferred to the client, normally over a network. Alternatively, the resulting video is included in a VOD library and made available to users on a VOD server.
• Referring toFIG. 19, a system may be developed that incorporates start detection of a play, end detection of a play, and summarization. The detection technique may be based upon processing a single frame, multiple frames, or a combination thereof.
• The terms and expressions which have been employed in the foregoing specification are used therein as terms of description and not of limitation, and there is no intention, in the use of such terms and expressions, of excluding equivalents of the features shown and described or portions thereof, it being recognized that the scope of the invention is defined and limited only by the claims which follow.

Claims (8)

1-54. (canceled)

55. A method of processing a video comprising:

(a) a service provider identifying a summary of said video;

(b) a service provider identifying a description of said summary; and

(c) said service provider providing said summary and said description to a remote user in response to receiving a request from said remote user.

56. The method ofclaim 55 wherein said user requests said summary with a browser.

57. A method of processing a video comprising:

(a) a service provider storing a summary of a video; and

(b) said service provider providing said summary to a remote user in response to receiving a request from said remote user.

58. The method ofclaim 57 wherein said user requests said summary with a browser.

59. A method of processing a video comprising:

(a) a service provider storing a description of a summary of a video; and

(b) said service provider providing said description to a remote user in response to receiving a request from said remote user.

60. The method ofclaim 59 wherein said user requests said description with a browser.

61-64. (canceled)

Images