VERFAHREN ZUR VIDEOANALYSE

• US 20120033083A1
• Filed: 05/19/2009
• Published: 02/09/2012
• Est. Priority Date: 05/28/2008
• Status: Active Grant

• Current Assignee: KiwiSecurity Software GmbH
• Original Assignee: KiwiSecurity Software GmbH
• Inventors: Hörbinger, Oliver, Kraus, Klemens, Röhr, Robert, Matusek, Florian, Sutor, Stephan
• Granted Patent: US 8,854,469 B2
• Time in Patent Office: Days
• US Class Current: 348/159
• CPC Class Codes: G06T 7/254 involving subtraction of images
• CPC Class Codes: H04N 7/181 for receiving images from a plurality of remote sources

Abstract

In a method for locating patterns of movement in a first digital video signal, wherein the first digital video signal comprising at least indirectly succeeding individual digital images is analyzed in real time, it is proposed for the locating of patterns of movement in a first digital video signal, during the occurrence of larger moving object volumes in real time, to assign at least a first location-changing foreground region in a first individual image in a predefined manner with a predefinable number of first markings, to subsequently determine the relative movement of the first markings between the first individual image and a subsequent second individual image, to subsequently associate each of the first markings with a predefinable first environment, to assign the first and/or at least one second location-changing foreground region in a predefined manner with a predefinable number of second markings, to remove the second markings disposed inside intersecting regions of a predefinable number of intersecting first environments, to subsequently determine the relative movement of the first and second markings between the second individual image and a subsequent third individual image, and to output the relative movements of the first and/or second markings as a first pattern of movement.

Claims

1-27. -27. (canceled)

28. A method for identifying movement patterns in a first digital video signal which comprises individual digital images arranged to at least indirectly in form a sequence, comprising the steps of:

analyzing the first digital video signal in real-time,

applying a predetermined number of first markings to at least one position-changing foreground region in a first individual image,

determining a relative movement of the first markings between the first individual image and a subsequent second individual image,

associating a predetermined first environment with each first marking in one-to-one correspondence,

applying a predetermined number of second markings to the first position-changing foreground region or to at least one second position-changing foreground region, or both,

removing the second markings arranged within intersecting regions of a predetermined number of intersecting first environments,

determining the relative movements of the first and second markings between the second individual image and a subsequent third individual image, and

outputting the relative movements of the first markings or the second markings, or both, as a first movement pattern.

29. The method of claim 28, wherein at least one of the first and second markings are arranged in form of a raster pattern.

30. The method of claim 29, wherein the raster pattern is uniformly distributed or distributed with a predetermined weighting.

31. The method of claim 28, wherein the relative movements of at least one of the first and second markings are determined by determining or estimating an optical flow in the first video signal.

32. The method of claim 31, wherein determining or estimating is performed with an optical-flow-tracker.

33. The method of claim 31, wherein a Horn-Schunk method or a Kanade-Lucas-Tomasi method.

34. The method of claim 28, wherein the first environment is formed as a rectangle, a square, or a sectional plane of a cone.

35. The method of claim 34, wherein the sectional plane of a cone is a circle or an ellipse.

36. The method of claim 34, wherein the first environment is weighted.

37. The method of claim 28, further comprising the steps of:

subdividing at least the second individual image into segments having a predetermined arrangement with a predetermined shape and size,

evaluating at least the first movement pattern within each segment, and

determining a segment movement pattern for each segment.

38. The method of claim 37, further comprising the step of determining at least one first main movement pattern from the first movement pattern or from the segment movement pattern, or both.

53. A computer program product embodied on a data carrier and including software code segments, which when loaded into a memory of a computer and executed on the computer, causes the computer to perform the method steps of claim 28.

61. The method of claim 28, wherein the at least one first position-changing foreground region is determined by:

recording a second individual image within a predetermined time subsequent to the first individual image,

subtracting the at least one background region of the first individual image from a background region of the second individual image, and

associating the so formed difference with the at least one first position-changing foreground region.

39. A method for identifying movement patterns in a first digital video signal which comprises individual digital images arranged to at least indirectly form a sequence, comprising the steps of:

analyzing the first digital video signal in real-time,

determining at least one background region in each individual image,

identifying at least one position-changing foreground region in relation to the background region, and

generating a message.

40. The method of claim 39, wherein the individual images have a predetermined number of pixels, and wherein each pixel has at least one of a grayscale value or a color value, the method further comprising the steps of:

comparing the value of a first pixel of a first individual image with a predetermined number of comparison values,

associating a value of a first comparison value from the predetermined number of comparison values with the first pixel if the value of the first pixel is within a predetermined tolerance range about the value of the first comparison value,

increasing a frequency of occurrence of the first comparison value by a predetermined first amount, and

associating the comparison value with the highest frequency of occurrence with the background region.

41. The method of claim 40, and further associating, if the value of the first pixel is located outside the predetermined tolerance range, the value of the first pixel with the value of a first comparison value having the lowest frequency of occurrence.

42. The method of claim 41, and further comprising the steps of:

recording a second individual image within a predetermined time subsequent to the first individual image,

subtracting the at least one background region of the first individual image from a background region of the second individual image, and

associating the so formed difference with at least the first position-changing foreground region.

43. The method of claim 40, and further reducing, after a predetermined number of subsequent individual images, a frequency of occurrence of each comparison value by a predetermined second amount.

44. The method of claim 41, and further comprising the steps of:

identifying at least one first position-changing object, which is part of the first position-changing foreground region, with a tracker, and

tracking movement of the at least one first position-changing object with the tracker.

45. The method of claim 44, wherein the tracker is a mean-shift-tracker or an optical-flow-tracker.

46. The method of claim 45, and further comprising the steps of:

selecting from a group of predetermined trackers a number of trackers for tracking the movement of the first position-changing object,

tracking the movement of the at least one first position-changing object with each of the selected trackers and determining a tracking result,

determining for each tracker an error probability, and

combining the tracking results from those trackers having the smallest error probabilities into a tracking end result.

47. The method of claim 45, and further comprising the steps of:

processing blocks comprising a predetermined number of pixels with a mean-shift-tracker by using an integral image, and

weighting individual blocks with at least one predetermined exponential function.

48. The method of claim 45, and further comprising the steps of:

processing blocks comprising a predetermined number of pixels with a mean-shift-tracker by using an integral image,

determining with the mean-shift-tracker for the individual blocks at least one vector for determining a center or a center of gravity of the at least one first position-changing object, with the at least one vector pointing in a direction of a region about the center or the center of gravity,

associating a predetermined second environment with each endpoint of the vector determined for each block,

arranging the respective second environments in a first intermediate image, and

applying the mean-shift-tracker to the first intermediate image.

49. The method of claim 39, and further comprising the steps of:

combining the first digital video signal and at least one other digital video signal, which are each associated with at least one partially overlapping recording region, to form a model of the recording region, and

at least one of outputting and displaying the model.

50. The method of claim 49, wherein the model is formed as at least one of a three-dimensional model or as a plan view model.

51. The method of claim 50, and further comprising the steps of:

displaying the plan view model and the three-dimensional model substantially simultaneously, and

selecting at least one of two digital video signals in the plan view model or in the three-dimensional model as video signal in the plan view model and in the three-dimensional model.

60. The security system of claim 39, further comprising transmitting the message to a control location.

52. Method for calibrating a pan-tilt-zoom (PTZ) camera, comprising the steps of:

orienting the PTZ camera sequentially towards an origin and towards three endpoints in three-dimensional space, wherein the origin and the three endpoints in conjunction form an orthogonal coordinate system with known distances between the origin and the three endpoints,

measuring, for each of the endpoints, three relative angles between the orientation to the origin and to an endpoint, and

determining from these relative angles a position of the PTZ camera in three-dimensional space with respect to the orthogonal coordinate system.

54. A video camera arrangement comprising at least one first video camera connected to a first data processing unit and at least one second video camera connected to a second data processing unit, said first and the second video camera recording at least partially an identical region and in conjunction with the first and second data processing units configured to identify at least one of a background region, a position-changing foreground region and a position-changing object.

55. The video camera arrangement of claim 54, wherein the first and second data processing units comprise at least one digital signal processor.

56. The video camera arrangement of claim 54, wherein the first and second data processing units are connected to a first central data processing unit configured to create a model of regions recorded by the first and the second video camera.

57. The video camera arrangement of claim 54, wherein the first central data processing unit is connected to a data store.

58. A security system comprising a video camera arrangement according to claim 54.

59. The security system of claim 58, further comprising a sensor selected from the group consisting of a light barrier, a motion detector, an impact sensor, a vibration sensor, a contact sensor, and an infrared sensor.