Reduction of video material to motion sections

Patent metadata

Field	Value
Publication number	US-2022180530-A1
Application number	US-202017440732-A
Country	US
Kind code	A1
Filing date	Mar 12, 2020
Priority date	Mar 19, 2019
Publication date	Jun 9, 2022
Grant date	—

Abstract

Systems and methods are provided that includes the processing of video material for reducing video material to temporal segments in which a significant movement of an object is recorded. The systems and methods may be used for the observation of animals.

First claim

1 . A method for reducing video recordings to temporal segments in which one or more movements of an object above a motion threshold are recorded, the method including steps comprising:
a) receiving a temporal sequence of images, b) generating a sequence of difference images from the sequence of images, by generating a respective difference image for each pair of adjacent images of the temporal sequence of images, c) generating a sequence of average-value difference images from the sequence of difference images, by a procedure in which, for all groups of successive difference images having a defined number of difference images, the difference images associated with a group are averaged in each case, d) generating a sequence of binary images from the sequence of average-value difference images, e) identifying groups of contiguous pixels in each binary image, f) determining the sizes of the groups of contiguous pixels and comparing the respective size of a group with a threshold value, g) identifying those binary images which have at least one group which is of a size equal to the threshold value or which is greater than the threshold value, and h) erasing all images of the temporal sequence of images which have not influenced the generation of a binary image identified in step g).

2 . The method as claimed in claim 1 ,
wherein each image of the sequence of images has a multiplicity of pixels, wherein each pixel is characterized by a tonal value, wherein each difference image of the sequence of difference images is characterized by a multiplicity of pixels each having a tonal value, wherein the tonal value of each pixel of each difference image represents an absolute value of the difference between the tonal values of the corresponding pixels of two temporally directly successive images, wherein a time window is defined for generating the sequence of average-value difference images, which time window can accommodate a defined number of temporally directly successive difference images, wherein the time window is shifted image by image from the beginning of the sequence of difference images until the end of the sequence of difference images, and, during each shifting image by image, a respective average-value difference image is generated on the basis of the difference images encompassed by the time window, wherein the tonal value of each pixel of each average-value difference image represents an average value of the tonal values of the corresponding pixels of the difference images encompassed by the time window, wherein, for generating the sequence of binary images from the sequence of average-value difference images, the tonal values of all pixels of each average-value difference image which lie below a defined tonal-value threshold value are set to a first tonal value, and the tonal values of all pixels of each (contrast-reduced) average-value difference image which lie above the defined tonal-value threshold value or correspond to the defined tonal-value threshold value are set to a second tonal value.

3 . The method as claimed in claim 1 , furthermore comprising the following step after step c) and before step d):
generating a sequence of contrast-reduced average-value difference images by applying a gaussian blur to all average-value difference images of the sequence of average-value difference images.

4 . The method as claimed in claim 1 , furthermore comprising the following step after step d) and before step e):
generating a sequence of dilated binary images from the sequence of binary images, by a procedure in which those pixels of each binary image which have the second tonal value are expanded singly or multiply to a shape of a defined structuring element.

5 . The method as claimed in claim 1 , wherein the size of each group in step f) is set as the number of pixels having a second tonal value which belong to the respective group.

6 . The method as claimed in claim 1 , wherein, for ascertaining the size of a group of contiguous pixels in a binary image, a bounding border around the group is considered which satisfies the following criteria:
the bounding border is rectangular, its edges run parallel to the edges of the binary image, all pixels which have the second tonal value and which belong to a group of contiguous pixels lie within the bounding border, the bounding border comprises as few pixels as possible which do not belong to the group of contiguous pixels having the second tonal value,
wherein the total number of pixels lying within the bounding border is set as the size of the group.

7 . The method as claimed in claim 1 , wherein the object is a living organism, preferably a living organism in the form of an animal.

8 . The method as claimed in claim 7 , wherein the at least one movement is shaking of the body of the living organism or of part of the body of the living organism, activities of the living organism concerning its own body care, and/or licking, chewing, scratching and/or rubbing of the living organism.

9 . A device comprising:
an input unit, a control unit, a computing unit, and an output unit and/or a data storage unit, wherein the control unit is configured to cause the input unit to receive a sequence of images, wherein the control unit is configured to cause the computing unit to carry out the following steps: a) generating a sequence of difference images from the sequence of images, by generating a respective difference image for each pair of adjacent images of the temporal sequence of images, b) generating a sequence of average-value difference images from the sequence of difference images, by a procedure in which, for all groups of successive difference images having a defined number of difference images, the difference images associated with a group are averaged in each case, c) generating a sequence of binary images from the sequence of average-value difference images, d) identifying groups of contiguous pixels in each binary image, e) determining the sizes of the groups of contiguous pixels and comparing the respective size with a threshold value, f) identifying those binary images which have at least one group which is of a size equal to the threshold value or which is greater than the threshold value, g) erasing all images of the temporal sequence of images which have not influenced the generation of a binary image identified in step f), wherein a reduced sequence of images is generated, wherein the control unit is configured to store the reduced sequence of images in the data storage unit and/or to cause the output unit to output the reduced sequence of images.

10 . The device as claimed in claim 9 , wherein the control unit is configured to cause the computing unit to:
a) receive a temporal sequence of images, b) generate a sequence of difference images from the sequence of images, by generating a respective difference image for each pair of adjacent images of the temporal sequence of images, c) generate a sequence of average-value difference images from the sequence of difference images, by a procedure in which, for all groups of successive difference images having a defined number of difference images, the difference images associated with a group are averaged in each case, d) generate a sequence of binary images from the sequence of average-value difference images, e) identify groups of contiguous pixels in each binary image, f) determine the sizes of the groups of contiguous pixels and comparing the respective size of a group with a threshold value, g) identify those binary images which have at least one group which is of a size equal to the threshold value or which is greater than the threshold value, and h

Assignees

• Bayer Animal Health Gmbh

Inventors

• Gillissen Philip

Classifications

• G06T7/254 Primary

  involving subtraction of images · CPC title

• G06T2207/20036

  Morphological image processing · CPC title

• G06V40/10

  Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands · CPC title

• G06T2207/10016

  Video; Image sequence · CPC title

• G06T7/215 Primary

  Motion-based segmentation · CPC title