Video processing system including advanced scene break detection methods for fades, dissolves and flashes

Abstract

The invention provides a method and apparatus for detecting a scene transition in digitized video data. The method includes determining if a gradual transition is detected in the video data that ends with a current frame based at least on the current frame and one or more previous frames in the digitized video data. The gradual transition has a length. The method includes comparing the gradual transition to a synthetically produced gradual transition spanning a same length and generated from starting and ending frames associated with the detected gradual transition and marking the detected gradual transition only if the detected gradual transition is sufficiently similar to the synthetically produced gradual transition.

1 . A method for detecting a scene transition in digitized video data comprising: acquiring digitized video data from a source; and evaluating a frame at a time the video data including for each frame 
 determining if a cut is detected between the current frame and a previous frame;    processing more frames in the digitized video data to determine if the detected cut is part of a long flash; and    marking detected cuts that are not long flashes.    
   
   
       2 . The method of  claim 1  wherein the step of processing more frames includes processing a predetermined number of frames including a last frame, the processing including comparing each frame including the last frame to the previous frame, and if any of the frames are sufficiently similar to the previous frame, then declaring the detected cut a long flash and not marking the detected cut.  
   
   
       3 . The method of claim I wherein comparing frames includes comparing a color difference between the frames.  
   
   
       4 . The method of  claim 1  wherein comparing frames includes comparing a motion difference.  
   
   
       5 . The method of  claim 1  wherein comparing the frames includes deriving a score representative of the similarity of the compared frames, the score derived from a mathematical comparison of the frames considering at least one parameter.  
   
   
       6 . The method of  claim 5  wherein the score is a product of the color and motion differences between the respective frames.  
   
   
       7 . The method of  claim 5  wherein the score is a weighted mean of the color and motion differences between the respective frames.  
   
   
       8 . A method for detecting a scene transition in digitized video data comprising: 
 determining if a dissolve is detected in the video data that ends with a current frame based at least on the current frame and one or more previous frames in the digitized video data, the dissolve having a length;    comparing the dissolve to a synthetically produced dissolve spanning a same length and generated from starting and ending frames associated with the detected dissolve; and    marking the detected dissolve only if the detected dissolve is sufficiently similar to the synthetically produced dissolve.    
   
   
       9 . The method of  claim 8  wherein the step of comparing includes comparing a subset of all of the frames in the detected dissolve to the frames in the synthetically produced dissolve.  
   
   
       10 . The method of  claim 8  wherein the step of determining further includes evaluating a previously processed frame at a distance a half dissolve from the current frame and a full dissolve from the current frame including determining if the current frame is sufficiently different from both the frame located a half dissolve length ago and a full dissolve length ago.  
   
   
       11 . The method of  claim 8  wherein the step of determining includes evaluating previously processed frames, a first frame at a distance a half dissolve from the current frame and a second frame at a distance a full dissolve from the current frame, including determining if the current frame is sufficiently different from both the first and second frames; and 
 marking the detected dissolve only if the current frame is sufficiently different from the first and second frames.    
   
   
       12 . The method of  claim 11  wherein prior to marking, the method further including 
 a. determining if the detected dissolve overlaps a previously detected dissolve;    b. if so, combining the overlapping dissolves.    
   
   
       13 . The method of  claim 8  wherein the determining step includes hypothesizing where the detected dissolve starts and ends including identifying starting, middle and ending frames; 
 comparing the starting, middle and ending frames to determine if each is sufficiently different from the others; and    if the frames are sufficiently different, marking the detected dissolve.    
   
   
       14 . The method of  claim 8  wherein comparing frames includes comparing a color difference between the frames.  
   
   
       15 . The method of  claim 8  wherein comparing frames includes comparing a motion difference.  
   
   
       16 . The method of  claim 8  wherein comparing the frames includes deriving a score representative of the similarity of the compared frames, the score derived from a mathematical comparison of the frames considering at least one parameter.  
   
   
       17 . The method of  claim 16  wherein the score is a product of the color and motion differences between the respective frames.  
   
   
       18 . The method of  claim 16  wherein the score is a weighted mean of the color and motion differences between the respective frames.  
   
   
       19 . A method for detecting a fade in digitized video data comprising: detecting a first transition in the digitized video data; 
 detecting a second transition within a pre-set number (Nfade) of frames from an end of the first transition;    determining if the frames intermediary between the first and second transitions are substantially uniform; and    marking the first and second transition as a single fade transition.    
   
   
       20 . The method of  claim 19  wherein the first transition is a dissolve.  
   
   
       21 . The method of  claim 19  wherein the second transition is a cut.  
   
   
       22 . The method of  claim 19  wherein the second transition is a dissolve.  
   
   
       23 . The method of  claim 19  wherein the step of comparing intermediary frames includes determining a number of frames between the first and second transitions, if the number of frames exceeds a predetermined number, marking each of the first and second transitions without marking a fade transition.  
   
   
       24 . The method of  claim 19  wherein the step of comparing intermediary frames includes determining if a minimum number of uniform frames exist between the first and second transitions, if the number of uniform frames does not exceed a predetermined number, marking each of the first and second transitions without marking a fade transition.  
   
   
       25 . The method of  claim 24  wherein the step of comparing intermediary frames includes determining a ratio of non-uniform frames to uniform frames between the first and second transitions, if the ratio exceeds a predetermined number, marking each of the first and second transitions without marking a fade transition.  
   
   
       26 . A method for detecting a compound transition in digitized video data comprising: detecting a first transition in the digitized video data; 
 detecting a second transition within a pre-set number (N fade ) of frames from the first transition;    determining if the frames intermediary between the first and second transitions are substantially uniform; and    marking the first and second transition as a compound transition.    
   
   
       27 . The method of  claim 26  wherein the compound transition is a fade.  
   
   
       28 . The method of  claim 27  wherein the first transition is a dissolve.  
   
   
       29 . The method of  claim 27  wherein the second transition is a cut or dissolve.  
   
   
       30 . The method of  claim 26  wherein comparing frames includes comparing a color difference between the frames.  
   
   
       31 . The method of  claim 26  wherein comparing frames includes comparing a motion difference.  
   
   
       32 . The method of  claim 26  wherein comparing the frames includes deriving a score representative of the similarity of the compared frames, the score derived from a mathematical comparison of the frames considering at least one parameter.  
   
   
       33 . The method of  claim 32  wherein the score is a product of the color and motion differences between the respective frames.  
   
   
       34 . The method of  claim 32  wherein the score is a weighted mean of the color and motion differences between the respective frames.  
   
   
       35 . The method of  claim 26  wherein prior to marking, the method further including 
 determining if the detected dissolve overlaps a previously detected dissolve; and    if so, combining the overlapping dissolves.    
   
   
       36 . A method for detecting a scene transition in digitized video data comprising: 
 determining if a gradual transition is detected in the video data that ends with a current frame based at least on the current frame and one or more previous frames in the digitized video data, the gradual transition having a length;    comparing the gradual transition to a synthetically produced gradual transition spanning a same length and generated from starting and ending frames associated with the detected gradual transition; and    marking the detected gradual transition only if the detected gradual transition is sufficiently similar to the synthetically produced gradual transition.    
   
   
       37 . The method of  claim 36  wherein the gradual transition is selected from the group of a dissolve, a wipe, a page flip, a tear, an iris transition and a pixelation transition.