Vehicle detecting system, vehicle detecting method, and program storage medium

Patent metadata

Field	Value
Publication number	US-12340575-B2
Application number	US-202318236772-A
Country	US
Kind code	B2
Filing date	Aug 22, 2023
Priority date	Jun 3, 2014
Publication date	Jun 24, 2025
Grant date	Jun 24, 2025

Abstract

Provided is an image processing system, an image processing method, and a program for preferably detecting a mobile object. The image processing system includes: an image input unit for receiving an input for some image frames having different times in a plurality of image frames constituting a picture, which is of a pixel on which the mobile object appears or a pixel on which the mobile object does not appear, for selected arbitrary one or more pixels in an image frame at the time of processing; and a mobile object detection model constructing unit for learning a parameter for detecting the mobile object based on the input.

First claim

The invention claimed is:

1. An image processing system comprising:
at least one memory storing instructions; and
at least one processor configured to process the instructions to control the image processing system to:
receive an input to place a plurality of first icons that correspond to a vehicle and to place a plurality of second icons that correspond to other than the vehicle on one or more pixels for each of a plurality of image frames comprising an image; and
generate learning data to train a detection model for detecting the vehicle based on the placement of the plurality of first icons and the plurality of second icons on the one or more pixels of the plurality of image frames,

wherein the at least one processor is configured to process the instructions to control the image processing system to:
receive an input to designate an object region or a background region with respect to one or more pixels at random positions in the plurality of image frames, wherein

the detection model is obtained by training three background models using the generated learning data, and these three background models are different from each other in the number of image frames to be processed,
a first background model has the largest number of image frames to be processed, a second background model has the second largest number of image frames to be processed, and a third background model has the smallest number of image frames to be processed,
the at least one processor is configured to process the instructions to control the image processing system to:
calculate a first distance between the first background model and the second background model, a second distance between the second background model and the third background model, and a third distance between the first background model and the third background model for each pixel or unit region of the plurality of image frames;
determine, based on the first distance, the second distance, and the third distance, whether or not the vehicle is shown in a region corresponding to each pixel of the plurality of image frames; and
based on user input specifying whether some pixels in the plurality of image frames are pixels that correspond the vehicle or a pixel that does not show the vehicle, learn parameters for detecting the vehicle in the plurality of image frames.

2. The image processing system according to claim 1 , wherein
the at least one processor is configured to process the instructions to control the image processing system to:
learn a parameter to be used for the detection model for detecting the vehicle by performing one or more convolution calculation by using a value of a background model of a neighboring region of the one or more pixels.

3. The image processing system according to claim 2 , wherein
the at least one processor is configured to process the instructions to control the image processing system to:
learn, as the parameter, a threshold compared with a value obtained as a result of the one or more convolution calculations.

4. The image processing system according to claim 1 , wherein
the first background model is generated based on the plurality of image frames,
wherein in the second background model, an influence of the plurality of image frames is smaller than that of the first background model, and
wherein in the third background model, an influence of the plurality of image frames is smaller than that of the second background mode.

5. An image processing method performed by at least one computer, the method comprising:
receiving an input to place a plurality of first icons that correspond to a vehicle and to place a plurality of second icons that correspond to other than the vehicle on one or more pixels for each of a plurality of image frames comprising an image; and generating learning data to train a detection model for detecting the vehicle based on the placement of the plurality of first icons and the plurality of second icons on the one or more pixels of the plurality of image frames,
wherein the at least one processor is configured to process the instructions to control the image processing system to:
receiving an input to designate an object region or a background region with respect to one or more pixels at random positions in the plurality of image frames, wherein

the detection model is obtained by training three background models using the generated learning data, and these three background models are different from each other in the number of image frames to be processed,
a first background model has the largest number of image frames to be processed, a second background model has the second largest number of image frames to be processed, and a third background model has the smallest number of image frames to be processed,
the method comprises:
calculating a first distance between the first background model and the second background model, a second distance between the second background model and the third background model, and a third distance between the first background model and the third background model for each pixel or unit region of the plurality of image frames;
determining, based on the first distance, the second distance, and the third distance, whether or not the vehicle is shown in a region corresponding to each pixel of the plurality of image frames; and
based on user input specifying whether some pixels in the plurality of image frames are pixels that correspond the vehicle or a pixel that does not show the vehicle, learning parameters for detecting the vehicle in the plurality of image frames.

6. The image processing method according to claim 5 , wherein the method comprises:
learning a parameter to be used for the detection model for detecting the vehicle by performing one or more convolution calculation by using a value of a background model of a neighboring region of the one or more pixels.

7. The image processing method according to claim 6 , wherein the method comprises:
learning, as the parameter, a threshold compared with a value obtained as a result of the one or more convolution calculations.

8. The image processing method according to claim 5 , wherein
the first background model is generated based on the plurality of image frames,
wherein in the second background model, an influence of the plurality of image frames is smaller than that of the first background model, and
wherein in the third background model, an influence of the plurality of image frames is smaller than that of the second background mode.

9. A non-transitory computer readable recording medium storing program instructions for causing a computer to perform:
receiving an input to place a plurality of first icons that correspond to a vehicle and to place a plurality of second icons that correspond to other than the vehicle on one or more pixels for each of a plurality of image frames comprising an image; and generating learning data to train a detection model for detecting the vehicle based on the placement of the plurality of first icons and the plurality of second icons on the one or more pixels of the plurality of image frames,
wherein the at least one processor is configured to process the instructions to control the image processing system to:
receiving an input to designate an object region or a background region with respect to one or more pixels at random positions in the plurality of image frames, wherein

the detection model is obtained by training three background models using the generated learning data, and these three background models are different from each other in the number of image frames to be processed,
a first background model has the largest number of image frames to be processed, a second b

Assignees

• Nec Corp

Inventors

• Miyano Hiroyoshi

Classifications

• G06T2207/10016

  Video; Image sequence · CPC title

• G06T7/70

  Determining position or orientation of objects or cameras (camera calibration G06T7/80) · CPC title

• G06V20/52

  Surveillance or monitoring of activities, e.g. for recognising suspicious objects (recognising microscopic objects G06V20/69) · CPC title

• G06V10/28

  Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns · CPC title

• G06T7/251

  involving models · CPC title