Image processing apparatus, image processing system, and image processing method
地域：
Tokyo
摘要
Provided are an apparatus and a method for executing a person detection process from an image. A divisional small region setting window is set and an image in the window is determined to be a person based on the number of moving body pixels of each divisional small region, and a first person candidate point is set in the window when the image is determined to be a person. A person determination process is executed based on an image feature quantity in a person determination detection frame, and a second person candidate point is set in the person determination detection frame when it is determined to be a person. A distribution state in a moving body detection frame of the person determination detection frame is analyzed, and an image in the person determination detection frame is determined to be a person based on an analysis result.
說明書
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
This application is a U.S. National Phase of International Patent Application No. PCT/JP2017/020052 filed on May 30, 2017, which claims priority benefit of Japanese Patent Application No. JP 2016-123245 filed in the Japan Patent Office on Jun. 22, 2016. Each of the above-referenced applications is hereby incorporated herein by reference in its entirety.
The present disclosure relates to an image processing apparatus, an image processing system, an image processing method, and a program. More particularly, the present disclosure relates to an image processing apparatus, an image processing system, an image processing method, and a program which are capable of executing a person detection process from image.
Recently, monitoring cameras have been installed in stations, buildings, public roads, and various other places. Captured images of the monitoring cameras are transmitted to a server via, for example, a network, stored in a storage device such as a database, and used for various types of data processing to which an information terminal connected to a server or a network is applied.
For example, data processing executed by the server or the information terminal is a process of searching for and tracking a specific person.
In a monitoring system using such a monitoring camera, various kinds of detection processes such as moving body detection, face detection, person detection, or the like are combined and used as a process for detecting a specific object such as a person from captured image data.
權(quán)利要求
The invention claimed is:1. An image processing apparatus, comprising:a data processing unit configured to execute a person detection process from an image, wherein, in the person detection process, the data processing unit is further configured to:execute a first person detection process to set a divisional small region setting window in a moving body detection frame surrounding a moving body region detected from the image; determine whether an image in the divisional small region setting window is a person based on a number of moving body pixels of each divisional small region; set a first person candidate point in the divisional small region setting window based on a determination that the image is the person; execute a second person detection process to set a person determination detection frame including the first person candidate point or a first person candidate point neighbor pixel; and determine whether an image in the person determination detection frame is the person based on an image feature quantity in the person determination detection frame. 2. The image processing apparatus according to claim 1 , wherein the data processing unit is further configured to:set a second person candidate point in the person determination detection frame based on the image in the person determination detection frame determined to be the person in the second person detection process; and execute a final determination process to analyze a distribution state in the moving body detection frame of the person determination detection frame in which the second person candidate point is set and finally determine whether the image in the person determination detection frame is the person based on an analysis result. 3. The image processing apparatus according to claim 1 , wherein, in the first person detection process, the data processing unit is further configured to:compare a proportion of the number of moving body pixels to a number of all pixels of each divisional small region with a threshold value specified in advance; and determine whether the image in the divisional small region setting window is the person based on a comparison result. 4. The image processing apparatus according to claim 1 , wherein the data processing unit is further configured to:generate a plurality of reduced images obtained based on reduction of the image with a plurality of different reduction rates; apply a non-reduced window of a same size to the plurality of generated reduced images; and execute the first person detection process. 5. The image processing apparatus according to claim 4 , whereinthe data processing unit is further configured to generate a first person candidate point list in which coordinate information is registered, and the coordinate information indicates a position of the first person candidate point obtained as a result of the first person detection process for a non-reduced original image and the plurality of reduced images obtained based on the reduction of the image with the plurality of different reduction rates. 6. The image processing apparatus according to claim 5 , whereinthe data processing unit is further configured to generate the first person candidate point list in which coordinate information in the non-reduced original image and information allowing a reduction level of an image to be identified are recorded as the coordinate information indicating the position of the first person candidate point, and a setting process for the first person candidate point is executed on the image. 7. The image processing apparatus according to claim 1 , wherein, in the second person detection process, the data processing unit is further configured to:calculate the image feature quantity in the person determination detection frame; execute a dictionary collation process which is a collation process of the calculated feature quantity and a pre-registered person determination feature quantity; and determine whether the image in the person determination detection frame is the person based on a collation result. 8. The image processing apparatus according to claim 1 , wherein, in the second person detection process, the data processing unit is further configured to:set a plurality of person determination detection frames centered on each pixel of a determined region having the first person candidate point as a center pixel; and determine whether the image in the person determination detection frame is the person based on the image feature quantity in each person determination detection frame of the plurality of person determination detection frames. 9. The image processing apparatus according to claim 1 , wherein the data processing unit is further configured to:generate a plurality of reduced images obtained based on reduction of the image with a plurality of different reduction rates; apply a non-reduced person detection process frame of a same size to the plurality of generated reduced images; and execute the second person detection process. 10. The image processing apparatus according to claim 9 , wherein the data processing unit is further configured to:generate a second person candidate point list in which coordinate information is registered, and the coordinate information indicates a position of a second person candidate point obtained as a result of the second person detection process for a non-reduced original image and the plurality of reduced images obtained based on the reduction of the image with the plurality of different reduction rates. 11. The image processing apparatus according to claim 10 , whereinthe data processing unit is further configured to generate the second person candidate point list in which coordinate information in the non-reduced original image and information allowing a reduction level of an image to be identified are recorded as the coordinate information indicating the position of the second person candidate point, and a setting process for the second person candidate point is executed on the image. 12. The image processing apparatus according to claim 2 , wherein, in the final determination process, the data processing unit is further configured to:calculate an occupancy rate of the person determination detection frame in which the second person candidate point is set in a specific region in the moving body detection frame; and finally determine that the image in the person determination detection frame is the person based on the occupancy rate that is equal to or greater than a threshold value. 13. The image processing apparatus according to claim 1 , whereinthe data processing unit is further configured to generate metadata in which object type information is recorded, and the object type information indicates whether a moving body obtained as a result of the second person detection process is the person. 14. The image processing apparatus according to claim 13 , further comprising a communication unit configured to transmit the metadata generated by the data processing unit through. 15. An image processing system, comprising:a camera configured to capture an image; and a server connected with the camera via a network, wherein the camera includes a data processing unit configured to execute a person detection process from the captured image, and in the person detection process, the data processing unit is further configured to:execute a first person detection process to set a divisional small region setting window in a moving body detection frame surrounding a moving body region detected from the captured image; determine whether an image in the divisional small region setting window is a person based on a number of moving body pixels of each divisional small region; set a first person candidate point in the divisional small region setting window based on a determination that the image is the person; execute a second person detection process to set a person determination detection frame including the first person candidate point or a first person candidate point neighbor pixel; determine whether an image in the person determination detection frame is the person based on an image feature quantity in the person determination detection frame; generate first metadata in which first object type information is recorded, wherein the first object type information indicates whether a moving body obtained as a result of the second person detection process is the person; and transmit the first metadata to the server. 16. The image processing system according to claim 15 , wherein the data processing unit of the camera is further configured to:set a second person candidate point in the person determination detection frame based on the image in the person determination detection frame determined to be the person in the second person detection process; execute a final determination process to analyze a distribution state in the moving body detection frame of the person determination detection frame in which the second person candidate point is set and finally determine whether the image in the person determination detection frame is the person based on an analysis result; and generate second metadata in which second object type information is recorded, wherein the second object type information indicates whether a moving body obtained as a result of the final determination process is the person; and transmit the second metadata to the server. 17. An image processing system, comprising:a camera configured to capture an image; and a server and an information terminal connected with the camera via a network, wherein the camera includes a data processing unit configured to execute a person detection process from the captured image, wherein, in the person detection process, the data processing unit is further configured to:execute a first person detection process to set a divisional small region setting window in a moving body detection frame surrounding a moving body region detected from the captured image; determine whether an image in the divisional small region setting window is a person based on a number of moving body pixels of each divisional small region; set a first person candidate point in the divisional small region setting window based on a determination that the image is the person; execute a second person detection process to set a person determination detection frame including the first person candidate point or a first person candidate point neighbor pixel; determine whether an image in the person determination detection frame is the person based on an image feature quantity in the person determination detection frame; generate first metadata in which first object type information is recorded, wherein the first object type information indicates whether a moving body obtained as a result of the second person detection process is the person; and transmit the first metadata to the server together with the captured image, and the server is configured to:store the captured image and the first metadata received from the camera in a storage unit; and execute an object search or a tracking process to which the captured image and the first metadata stored in the storage unit are applied based on a processing request from the information terminal. 18. The image processing system according to claim 17 , wherein the data processing unit of the camera is further configured to:set a second person candidate point in the person determination detection frame based on the image in the person determination detection frame determined to be the person in the second person detection process; execute a final determination process to analyze a distribution state in the moving body detection frame of the person determination detection frame in which the second person candidate point is set and finally determine whether the image in the person determination detection frame is the person based on an analysis result; and generate second metadata in which second object type information is recorded, wherein the second object type information indicates whether a moving body obtained as a result of the final determination process is the person; and transmit the second metadata to the server. 19. An image processing method, comprising:executing, by a data processing unit of an image processing apparatus, a first person detection process of setting a divisional small region setting window in a moving body detection frame surrounding a moving body region detected from an image; determining, by the data processing unit, whether an image in the divisional small region setting window is a person based on a number of moving body pixels of each divisional small region; setting, by the data processing unit, a first person candidate point in the divisional small region setting window based on the determination that the image is the person; executing, by the data processing unit, a second person detection process of setting a person determination detection frame including the first person candidate point or a first person candidate point neighbor pixel; and determining whether an image in the person determination detection frame is the person based on an image feature quantity in the person determination detection frame. 20. A non-transitory computer-readable medium having stored thereon, computer-executable instructions which, when executed by an image processing apparatus, cause the image processing apparatus to execute operations, the operations comprising:executing, by a data processing unit of the image processing apparatus, a first person detection process of setting a divisional small region setting window in a moving body detection frame surrounding a moving body region detected from an image; determining, by the data processing unit, whether an image in the divisional small region setting window is a person based on number of moving body pixels of each divisional small region; setting, by the data processing unit, a first person candidate point in the divisional small region setting window based on the determination that the image is the person; executing, by the data processing unit, a second person detection process of setting a person determination detection frame including the first person candidate point or a first person candidate point neighbor pixel; and determining, by the data processing unit, whether an image in the person determination detection frame is the person based on an image feature quantity in the person determination detection frame.
意見反饋
使用該功能遇到問題