# US 10,872,423 B2 · Image detection device, image detection method and storage medium storing program

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 10,872,423 B2
- Document ID: US10872423B2
- Status: Granted
- Kind: B2
- Application number: 16/561,444
- Filed: 2019-09-05
- Published: 2020-01-02
- Granted: 2020-12-22
- Earliest claimed: 2016-03-31
- Assignees: NEC Corporation
- Applicants: NEC Corporation
- Inventors: Takuya Ogawa
- Web page: https://patentgrep.com/patent/US10872423B2

## CPC

- H04N 7/183: {for receiving images from a single remote source}
- G06T 7/00: Image analysis
- G06T 7/20: Analysis of motion (motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51)
- G06T 7/246: using feature-based methods, e.g. the tracking of corners or segments
- G06T 7/248: {involving reference images or patches}
- G06V 10/255: {Detecting or recognising potential candidate objects based on visual cues, e.g. shapes}
- G06V 10/772: Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries
- G08B 13/19617: {Surveillance camera constructional details}
- H04N 7/18: Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- G06T 2207/30241: Trajectory
- G06V 10/50: by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
- G06V 10/62: relating to a temporal dimension, e.g. time-based feature extraction; Pattern tracking
- G06V 10/759: {Region-based matching}
- G06V 2201/08: Detecting or categorising vehicles
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G08B: SIGNALLING SYSTEMS, e.g. PERSONAL CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS

## Abstract

Provided are an image detection device, an image detection method and a program, which are capable of improving correspondence to a target deformation by optimizing a template shape, when performing target detection using template matching. An image detection device 100 for detecting a target from an input image comprises: a template generation unit 10 that generates a template for detecting a target; a mask generation unit 20 that generates a mask which shields a portion of the template, on the basis of temporal variations of a feature point extracted from an area including the image target; and a detection unit 30 that detects the target from the image using the template a portion of which is shielded by the mask.

## Claims

### Claim 1

An image detection device comprising:
at least one memory that stores a set of instructions; and
at least one processor configured to execute the set of instructions to:
generate a template for detecting a target;
generate a mask partly shielding the template based on temporal variation of feature points extracted from an area including the target in an input image; and
detect the target from the image by using the template partly shielded by the mask, wherein
the at least one processor is further configured to:
determine, when updating the mask, whether a count of subregions not to be shielded is equal to or larger than a predetermined value, the subregions being regions into which an area including the target in an image;
generate the mask when the count of the subregions not to be shielded by the mask is equal to or larger than the predetermined value; and
cause the mask to shield the template, and
the at least one processor is configured not to generate the mask when the count of the subregions to be shielded by the mask is larger than the predetermined value.

### Claim 2 (depends on 1)

The image detection device according to claim 1, wherein
the at least one processor is further configured to:
calculate moving directions to second feature points in a second image from the respective feature points, the second feature points corresponding to the respective feature points;
determine at most frequent moving direction among the moving directions;
determine whether occurring number of the most frequent moving direction is equal to or larger than a half of a total number of the moving directions;
count the second feature points moving in the most frequent direction in each of the subregions when the occurring number is not equal to or larger than the total number;
count the second feature points in each of the subregions when the occurring number is equal to or larger than the total number; and
generate the mask based on a counted number of the second feature points in each of the subregions.

### Claim 3 (depends on 1)

The image detection device according to claim 1, wherein
the at least one processor is further configured to:
identify, in a second image, second feature points corresponding to the feature points, and
identify, as the area including the target, a rectangle enclosing regions in which the second feature points are distributed.

### Claim 4 (depends on 1)

The image detection device according to claim 1, wherein
the at least one processor is further configured to:
divide an area in an input image into subregions, the area including the target; and
extract centers of the subregions as feature points.

### Claim 5

An image detection method comprising:
generating a template for detecting a target;
generating a mask partly shielding the template based on temporal variation of feature points extracted from an area including the target in an input image; and
detecting the target from the image by using the template partly shielded by the mask, wherein
the method further comprises:
determining, when updating the mask, whether a count of subregions not to be shielded is equal to or larger than a predetermined value, the subregions being regions into which an area including the target in an image;
generating the mask when the count of the subregions not to be shielded by the mask is equal to or larger than the predetermined value;
causing the mask to shield the template; and
generating no mask when the count of the subregions to be shielded by the mask is larger than the predetermined value.

### Claim 6 (depends on 5)

The image detection method according to claim 5, the method further comprising:
calculating moving directions to second feature points in a second image from the respective feature points, the second feature points corresponding to the respective feature points;
determining a most frequent moving direction among the moving directions;
determining whether occurring number of the most frequent moving direction is equal to or larger than a half of a total number of the moving directions;
counting the second feature points moving in the most frequent direction in each of the subregions when the occurring number is not equal to or larger than the total number;
counting the second feature points in each of the subregions when the occurring number is equal to or larger than the total number; and
generating the mask based on a counted number of the second feature points in each of the subregions.

### Claim 7 (depends on 5)

The image detection method according to claim 5, the method further comprising:
identifying, in a second image, second feature points corresponding to the feature points; and
identifying, as the area including the target, a rectangle enclosing regions in which the second feature points are distributed.

### Claim 8

A non-transitory computer readable storage medium storing a program causing a computer to execute:
template generating processing of generating a template for detecting a target;
mask generation processing of generating a mask partly shielding the template based on temporal variation of feature points extracted from an area including the target in an input image; and
detection processing of detecting the target from the image by using the template partly shielded by the mask, wherein
the mask generation processing includes:
processing of determining, when updating the mask whether a count of subregions not to be shielded is equal to or larger than a predetermined value, the subregions being regions into which an area including the target in an image;
processing of generating the mask when the count of the subregions not to be shielded by the mask is equal to or larger than the predetermined value; and
processing of causing the mask to shield the template, and
the mask generation processing do not generate the mask when the count of the subregions to be shielded by the mask is larger than the predetermined value.

### Claim 9 (depends on 8)

The non-transitory computer readable storage medium according to claim 8, wherein
the mask generation processing includes:
processing of calculating moving directions to second feature points in a second image from the respective feature points, the second feature points corresponding to the respective feature points;
processing of determining a most frequent moving direction among the moving directions;
processing of determining whether occurring number of the most frequent moving direction is equal to or larger than a half of a total number of the moving directions;
processing of counting the second feature points moving in the most frequent direction in each of the subregions when the occurring number is not equal to or larger than the total number;
processing of counting the second feature points in each of the subregions when the occurring number is equal to or larger than the total number; and
processing of generating the mask based on a counted number of the second feature points in each of the subregions.

### Claim 10 (depends on 8)

The non-transitory computer readable storage medium according to claim 8, wherein
the tracking processing includes:
processing of identifying, in a second image, second feature points corresponding to the feature points; and
processing of identifying, as the area including the target, a rectangle enclosing regions in which the second feature points are distributed.

### Claim 11 (depends on 8)

The non-transitory computer readable storage medium according to claim 8, wherein
the program causes a computer to further execute:
tracking processing of dividing an area in an input image into subregions, the area including the target; and
extractions processing of extracting centers of the subregions as feature points.

### Claim 12 (depends on 5)

The image detection method according to claim 5, the method further comprises:
dividing an area in an input image into subregions, the area including the target; and
extracting centers of the subregions as feature points.