Back to results
US 11,386,536 B2Granted
Image analysis device, image analysis method and image analysis program
AssigneeNEC Corporation
InventorTakeshi Arikuma
Life of this patent
Estimating the term…
Abstract
The present invention provides an image analysis device which is capable of automatically responding to various environmental variations caused by a camera installation condition or an environmental factor without consuming unnecessary calculation resources. The image analysis device is provided with: a plurality of process execution units which are capable of executing different processes on an input image; an analysis unit which analyzes, on the basis of the image, an image variation caused by external environment; and a process selection unit which selects, on the basis of the analyzed variation, at least one from among the plurality of process execution units.
Claims18
An image analysis device comprising:
- at least one memory storing a set of instructions;
- and
- at least one processor configured to execute the set of instructions to:
- detect a plurality of faces from a plurality of frames included in an input video;
- select sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms;
- sort a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- output the sorted face,
- wherein
- the plurality of sorting algorithms include a plural output filter configured to select a plurality of faces having best quality in blur from a plurality of faces of a same person, and
- the at least one processor is further configured to execute the set of instructions to:
- execute processing by a previous filter and processing by the plural output filter and output a face selected by the previous filter until the transition time period elapses when the plural output filter is selected, the previous filter having been selected from the plurality of sorting algorithms when the plural output filter is selected;
- and
- stop the processing by the previous filter and output a plurality of faces selected by the plural output filter from a detected plurality of faces when the transition time period has elapsed.
11 more elements
An image analysis device comprising:
- at least one memory storing a set of instructions;
- and
- at least one processor configured to execute the set of instructions to:
- detect a plurality of faces from a plurality of frames included in an input video;
- extract environment information from the input video;
- select sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sort a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- output the sorted face,
- wherein
- the plurality of sorting algorithms include a single output filter configured to select one face having best quality in a plurality of faces of a same person, the plurality of faces being detected during a past predetermined time period,
- the environment information includes brightness bias, and
- the at least one processor is further configured to execute the set of instructions to
- select the single output filter when the brightness bias is larger than a brightness threshold.
11 more elements
An image analysis device comprising:
- at least one memory storing a set of instructions;
- and
- at least one processor configured to execute the set of instructions to:
- detect a plurality of faces from a plurality of frames included in an input video;
- extract environment information from the input video;
- select sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sort a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- output the sorted face,
- wherein
- the plurality of sorting algorithms include a plural output filter configured to select a plurality of faces having best quality in blur from a detected plurality of faces of a same person,
- the environment information includes a blurring amount, and
- the at least one processor is further configured to execute the instructions to
- select the plural output filter when the blurring amount is larger than a blur threshold.
11 more elements
An image analysis device comprising:
- at least one memory storing a set of instructions;
- and
- at least one processor configured to execute the set of instructions to:
- detect a plurality of faces from a plurality of frames included in an input video;
- extract environment information from the input video;
- select sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sort a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- output the sorted face,
- wherein
- the plurality of sorting algorithms include a simple selection filter configured to select a face having quality higher than a quality threshold,
- the environment information includes brightness bias and a blurring amount, and
- the at least one processor is further configured to execute the set of instructions to
- select the simple selection filter when the brightness bias is less than a brightness threshold and the blurring amount is less than a blur threshold.
11 more elements
An image analysis method comprising:
- detecting a plurality of faces from a plurality of frames included in an input video;
- selecting sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms;
- sorting a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- outputting the sorted face, wherein
- the plurality of sorting algorithms include a plural output filter configured to select a plurality of faces having best quality in blur from a plurality of faces of a same person, and
- the image analysis method further comprising:
- executing processing by a previous filter and processing by the plural output filter and outputting a face selected by the previous filter until the transition time period elapses when the plural output filter is selected, the previous filter having been selected from the plurality of sorting algorithms when the plural output filter is selected;
- and
- stopping the processing by the previous filter and outputting a plurality of faces selected by the plural output filter from a detected plurality of faces when the transition time period has elapsed.
7 more elements
An image analysis method comprising:
- detecting a plurality of faces from a plurality of frames included in an input video;
- extracting environment information from the input video;
- selecting sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sorting a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- outputting the sorted face, wherein
- the plurality of sorting algorithms include a single output filter configured to select one face having best quality in a plurality of faces of a same person, the plurality of faces being detected during a past predetermined time period,
- the environment information includes brightness bias, and
- the selecting includes selecting the single output filter when the brightness bias is larger than a brightness threshold.
6 more elements
An image analysis method comprising:
- detecting a plurality of faces from a plurality of frames included in an input video;
- extracting environment information from the input video;
- selecting sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sorting a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- outputting the sorted face, wherein
- the plurality of sorting algorithms include a plural output filter configured to select a plurality of faces having best quality in blur from a detected plurality of faces of a same person,
- the environment information includes a blurring amount, and
- the selecting includes selecting the plural output filter when the blurring amount is larger than a blur threshold.
6 more elements
An image analysis method comprising:
- detecting a plurality of faces from a plurality of frames included in an input video;
- extracting environment information from the input video;
- selecting sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sorting a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- outputting the sorted face, wherein
- the plurality of sorting algorithms includes a simple selection filter configured to select a face having quality higher than a quality threshold,
- the environment information include brightness bias and a blurring amount, and
- the selecting includes selecting the simple selection filter when the brightness bias is less than a brightness threshold and the blurring amount is less than a blur threshold.
6 more elements
A non-transitory computer readable storage medium storing a program causing a computer to execute:
- detection processing of detecting a plurality of faces from a plurality of frames included in an input video;
- algorithm selection processing of selecting sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms;
- sorting processing of sorting a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- output processing of outputting the sorted face, wherein
- the plurality of sorting algorithms include a plural output filter configured to select a plurality of faces having best quality in blur from a plurality of faces of a same person, and
- the program further causing the computer to execute:
- processing of executing processing by a previous filter and processing by the plural output filter and of outputting a face selected by the previous filter until the transition time period elapses when the plural output filter is selected, the previous filter having been selected from the plurality of sorting algorithms when the plural output filter is selected;
- and
- processing of stopping the processing by the previous filter and of outputting a plurality of faces selected by the plural output filter from a detected plurality of faces when the transition time period has elapsed.
7 more elements
A non-transitory computer readable storage medium storing a program causing a computer to execute:
- detection processing of detecting a plurality of faces from a plurality of frames included in an input video;
- environment extraction processing of extracting environment information from the input video;
- algorithm selection processing of selecting sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sorting processing of sorting a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- output processing of outputting the sorted face, wherein
- the plurality of sorting algorithms include a single output filter configured to select one face having best quality in a plurality of faces of a same person, the plurality of faces being detected during a past predetermined time period,
- the environment information includes brightness bias, and wherein
- the algorithm selection processing includes processing of selecting the single output filter when the brightness bias is larger than a brightness threshold.
6 more elements
A non-transitory computer readable storage medium storing a program causing a computer to execute:
- detection processing of detecting a plurality of faces from a plurality of frames included in an input video;
- environment extraction processing of extracting environment information from the input video;
- algorithm selection processing of selecting sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sorting processing of sorting a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- output processing of outputting the sorted face, wherein
- the plurality of sorting algorithms include a plural output filter configured to select a plurality of faces having best quality in blur from a detected plurality of faces of a same person,
- the environment information includes a blurring amount, and wherein
- the algorithm selection processing includes processing of selecting the plural output filter when the blurring amount is larger than a blur threshold.
6 more elements
A non-transitory computer readable storage medium storing a program causing a computer to execute:
- detection processing of detecting a plurality of faces from a plurality of frames included in an input video;
- environment extraction processing of extracting environment information from the input video;
- algorithm selection processing of selecting sorting algorithms from a plurality of sorting algorithms different from each other based on characteristics of the plurality of sorting algorithms and the environment information;
- sorting processing of sorting a face having a high quality from the detected faces according to the selected sorting algorithms in consideration of a transition time period after which a sorting algorithm produces stable results;
- and
- output processing of outputting the sorted face, wherein
- the plurality of sorting algorithms include a simple selection filter configured to select a face having quality higher than a quality threshold,
- the environment information includes brightness bias and a blurring amount, and wherein
- the algorithm selection processing includes processing of selecting the simple selection filter when the brightness bias is less than a brightness threshold and the blurring amount is less than a blur threshold.
6 more elements
Dependencies are inferred from the claim text.