Image analysis system, image analysis method, and storage medium

• US 10,552,713 B2
• Filed: 02/05/2015
• Issued: 02/04/2020
• Est. Priority Date: 04/28/2014
• Status: Active Grant

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Koyama, Kazuya
• Primary Examiner(s): Osinski, Michael S
• Application Number:
  US15/307,600
• Publication Number:
  US 20170053191A1
• Time in Patent Office: 1,825 Days
• CPC Class Codes: G06F 18/24 Classification techniques
• CPC Class Codes: G06V 20/52 Surveillance or monitoring of activities, e.g. for recognising suspicious objects recognising microscopic objects G06V20/69
• CPC Class Codes: G06V 40/173 face re-identification, e.g. recognising unknown faces across different face tracks
• CPC Class Codes: G08B 25/04 using a single signalling line, e.g. in a closed loop

Abstract

A monitor target shooting information generation unit extracts, from each of a plurality of images, as information to be used for estimating the identify of a monitor target, monitor target identification information that is identification information of the monitor target, and the monitor target shooting information generation unit then generates multiple pieces of monitor target shooting information each including both the extracted monitor target identification information and a shooting time at which the monitor target was shot. An appearance history generation unit generates, from the generated multiple pieces of monitor target shooting information, an appearance history of the monitor target that has been estimated to be identical. A determination unit determines the monitor target the appearance history of which matches a specified rule.

Claims

1. An image analysis system comprising:

a memory storing instructions;

a central processing unit (CPU) connected to the memory that, based on the instructions, performs operations comprising;

generating an appearance record of an object, the appearance record representing a time when the object appeared in a camera captured image and a location of the object when the object appeared in the camera captured image, the appearance record including appearance time information and location information of the object specified by the camera captured image, wherein generating the appearance record comprises determining that a continuous appearance of the object in one or more camera captured images during a predetermined period constitutes one appearance;

specifying the object as a monitored object when both

a time interval is within a predetermined time interval, the time interval being calculated using the time information in the appearance record, and

an appearance pattern in the time interval is a predetermined pattern, the appearance pattern being specified based on the location information in the appearance record; and

providing output identifying the monitored object.

2. The image analysis system according to claim 1, the operations further comprising identifying the monitored object based on similarity between appearance records of a plurality of objects.

3. The image analysis system according to claim 2,

wherein identifying the monitored object further comprises determining whether the plurality of objects match a rule based on a similarity between the appearance records of the plurality of objects.

6. An image analysis system according to claim 1, the operations further comprising:

detecting an event, the predetermined time interval being determined based on a time when the detected event occurs.

7. An image analysis system according to claim 6, the operations further comprising:

detecting an abnormal sound detected by a microphone as the event.

8. An image analysis system according to claim 6, the operations further comprising:

detecting smoke detected by a smoke detector as the event.

9. An image analysis system according to claim 6, the operations further comprising:

detecting the event based on information existing on the internet.

10. An image analysis system according to claim 6, the operations further comprising:

predicting the time and place where a crime occurs; and

detecting the predicted time and place as the event.

11. An image analysis system according to claim 6, wherein the predetermined pattern is determined based on a location of the detected event.

4. An image analysis method comprising:

generating an appearance record of an object, the appearance record representing a time when the object appeared in a camera captured image and a location of the object when the object appeared in the camera captured image, the appearance record including appearance time information and location information of the object specified by the camera captured image, wherein generating the appearance record comprises determining that a continuous appearance of the object in one or more camera captured images during a predetermined period constitutes one appearance;

specifying the object as a monitored object when both

a time interval is within a predetermined time interval, the time interval being calculated using the time information in the appearance record, and

an appearance pattern in the time interval is a predetermined pattern, the appearance pattern being specified based on the location information in the appearance record; and

providing output identifying the monitored object.

5. A non-transitory storage medium storing an analysis program including a processor causing a computer to execute:

generating an appearance record of an object, the appearance record representing a time that the object appeared in a camera captured image and a location of the object when the object appeared in the camera captured image, the appearance record including appearance time information and location information of the object specified by the camera captured image, wherein generating the appearance record comprises determining that a continuous appearance of the object in one or more camera captured images during a predetermined period constitutes one appearance;

specifying the object as a monitored object when both

a time interval is within a predetermined time interval, the time interval being calculated using the time information in the appearance record, and

an appearance pattern in the time interval is a predetermined pattern, the appearance pattern being specified based on the location information in the appearance record; and

providing output identifying the monitored object.