KR20140144977A - Video analysis optimization apparatus and method for surveillance site - Google Patents

Video analysis optimization apparatus and method for surveillance site Download PDF

Info

Publication number
KR20140144977A
KR20140144977A KR1020130067279A KR20130067279A KR20140144977A KR 20140144977 A KR20140144977 A KR 20140144977A KR 1020130067279 A KR1020130067279 A KR 1020130067279A KR 20130067279 A KR20130067279 A KR 20130067279A KR 20140144977 A KR20140144977 A KR 20140144977A
Authority
KR
South Korea
Prior art keywords
image
image analysis
ground
optimization
algorithm
Prior art date
Application number
KR1020130067279A
Other languages
Korean (ko)
Inventor
이원경
Original Assignee
이원경
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 이원경 filed Critical 이원경
Priority to KR1020130067279A priority Critical patent/KR20140144977A/en
Publication of KR20140144977A publication Critical patent/KR20140144977A/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30232—Surveillance

Abstract

The present invention relates to an apparatus and a method for optimizing video analysis for a surveillance site. More specifically, an in-situ image photographed from a surveillance camera is received to apply a plurality of image analysis algorithms to the image analysis of the in-situ image for the test of the image analysis algorithm, thereby comparing an image analysis result value from the test with a result image of a ground truth image to generate an optimized image analysis algorithm. Accordingly, a surveillance environment optimized for each surveillance site can be easily constructed regardless of the help of an expert. Information of parameters for the image analysis algorithms and the information of a weighted value are adjusted for the test of the algorithms, thereby creating the image analysis algorithm for the surveillance site including a plurality of image analysis algorithms and having parameters and weighted values through the comparison with the ground truth image so that the image analysis algorithms can be easily applied to an individual surveillance site. The setting information of the image analysis algorithm optimized according to periods and times through the ground truth image created based on a period, a time, or accumulation information is applied, so that the optimized image analysis algorithm for the surveillance site optimized for a surveillance environment is applied, thereby improving the efficiency in the construction of the surveillance environment for the surveillance site.

Description

TECHNICAL FIELD [0001] The present invention relates to an image analysis optimization apparatus and method for a surveillance site,

The present invention relates to an image analysis optimization apparatus and method for a surveillance site, and more particularly, to an image analysis optimization apparatus and method for a surveillance site that optimizes an image analysis efficiency of a shot image on a surveillance site (site).

There is a disadvantage that a conventional video analysis configuration in which a scene image taken through a surveillance camera installed in a surveillance site is input and performs an image analysis is difficult to change after being set once.

As a result, if the image environment changes due to noise, illuminance, weather, meaningless data such as shaking branches or waves, the error rate of image analysis increases and the efficiency of surveillance drops sharply.

Therefore, in the surveillance system having the conventional image analysis configuration, the scene environment in which the surveillance camera is installed can be manually set manually in response to the changing specific environment (for example, temporal change such as daytime and nighttime or environmental or seasonal change) , The efficiency of the image analysis was maintained.

However, such optimization of image analysis is difficult for general user to perform work because of manual adjustment or complex setting, and it is difficult for general users to perform the work. The maintenance cost has been consumed a lot.

In addition, even if the expert performs the optimization process, since it can apply the appropriate image analysis algorithm to the monitoring environment through the long-time image analysis algorithm test on the field image, the time burden for maintenance is considerably high.

In addition, when the monitoring environment changes gradually due to temporal, environmental, and seasonal changes, it is necessary to closely observe it and change it to an appropriate image analysis algorithm. Therefore, the effectiveness of the image analysis remains constant regardless of changes in the monitoring environment There was a disadvantage that it could not.

Korean Patent No. 10-0980586 [Title: Intelligent Image Security Security Method and System Using Single or Multiple Cameras]

According to an aspect of the present invention, there is provided a method of analyzing a plurality of image analysis algorithms by applying a plurality of image analysis algorithms to an image analysis of a field image by receiving a field image taken from a surveillance camera, The present invention provides an image analysis optimization apparatus and method for a surveillance site that generates an optimized image analysis algorithm in comparison with a result of an image analysis.

It is another object of the present invention to improve the above-mentioned problems by adjusting and testing parameter information and weight information for a plurality of image analysis algorithms, thereby obtaining optimized parameters including a plurality of image analysis algorithms And an image analysis optimization apparatus and method for a surveillance site that generates an image analysis algorithm for a surveillance site having a weight.

Another object of the present invention to improve the above-mentioned problems is to provide an image analysis algorithm optimization information that is optimized for each cycle and time by comparing the image with the ground actual image generated based on the period, time, or cumulative information The present invention provides an image analysis optimization apparatus and method for a surveillance site that is adapted to an image analysis algorithm for a surveillance site that is optimized for a surveillance environment that varies according to a time zone of an individual surveillance site.

Another object of the present invention to overcome the above problems is to provide an image analysis algorithm for each season such as early summer, late autumn, and snow, rain, and fog, And to provide an image analysis optimization apparatus and method for a surveillance site to which an image analysis algorithm for an optimized surveillance site is applied.

Another object of the present invention to solve the above problems is to provide a video surveillance system that can be configured in various physical locations such as a surveillance camera, a DVR server, an NVR server, and a central control server, And to provide an image analysis optimization apparatus and method for a monitoring site.

According to another aspect of the present invention, there is provided an image analysis optimization apparatus for a surveillance site, comprising: a field image taken from a surveillance camera; at least one of parameter information and weight information is adjusted; Analyzing the analysis algorithm to image analysis of the field image and comparing the image analysis result of the test with the result of the ground truth image of the field image to obtain optimized parameter information and weight information An optimization image analysis algorithm is generated and the optimization image analysis algorithm having the parameter information and the weight information optimized for image analysis of the field image input thereafter is applied.

According to another aspect of the present invention, there is provided an apparatus for optimizing an image analysis for a surveillance site, the system comprising: a field image input unit for receiving a field image taken from a surveillance camera; a ground truth image for the field image; An image analysis algorithm management unit that manages respective setting information for a plurality of image analysis algorithms, and an image analysis algorithm management unit that applies the plurality of image analysis algorithms to image analysis of the field image, An image analysis algorithm testing section for adjusting and testing information of the field image, and an image analyzing section for comparing the image analysis result of the test with the result of the ground image to generate an optimized image analysis algorithm having the setting information optimized An image matching unit, The image analysis of the image comprises image analysis unit optimize the application of the setting information of the optimized image analysis algorithms.

The setting information preferably includes at least one of parameter information of each image analysis algorithm and weight information of each image analysis algorithm.

Wherein the ground-based image generating unit generates the ground-based real-time image of the field-of-interest image periodically input for a predetermined test time on a time-by-time basis, The setting information of the optimization image analysis algorithm can be applied in consideration of the time.

In addition, in the comparison with respect to the field image, the ground-side actual image matching unit may include an optimization image having the setting information optimized and compared based on at least one of cumulative information of an image analysis result of the test and a result of the ground- You can also create an analysis algorithm.

Alternatively, the ground-based image generating unit may generate the ground-based actual image of the input scene image for each environment according to a preset reference, and the image analysis optimizing unit may perform an image analysis on the input scene image, And the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each environment may be applied.

Alternatively, the ground-based image generation unit may generate the ground-based actual image for each environment according to a criterion including at least one of snow, rain, fog, and yellow dust.

Alternatively, the ground-based image generating unit may generate the ground-based actual image for the input scene image, for each season, and the image analysis optimizing unit may generate the ground image based on the seasonally generated image analysis, The setting information of the optimization image analysis algorithm generated through the comparison of the test result values may be applied.

Meanwhile, the image analysis optimization apparatus for a surveillance site according to another embodiment of the present invention may further include parameter information and weight information of the optimization image analysis algorithm for the scene image compared with the periodically generated ground image, And may further include an optimization setting management unit that classifies and manages the data by the period.

In addition, the image analysis optimization apparatus for a surveillance site according to another embodiment of the present invention may physically be located in one of a surveillance camera, a DVR server, an NVR server, and a central control server.

According to an aspect of the present invention, there is provided an image analysis optimization method for a surveillance site, the method comprising: receiving a scene image taken from a surveillance camera through a scene image input unit; Generating a ground truth image for a scene image, managing each setting information for a plurality of image analysis algorithms through an image analysis algorithm management unit, Applying an image analysis algorithm to the image analysis of the field image, adjusting and testing each of the setting information, and transmitting the image analysis result of the test to the ground image And the optimization with the setting information Generating an image analysis algorithm, and applying the setting information of the optimization image analysis algorithm to an image analysis of the field image inputted through the image analysis optimization unit.

The method of optimizing an image analysis for a surveillance site according to an exemplary embodiment of the present invention includes optimizing setting management unit for generating parameter information and weight information of the optimized image analysis algorithm for the field image compared with the periodically- As the setting information, classifying and managing the information according to the period.

Wherein the ground-based image generating unit generates the ground-based real-time image for the field-of-interest periodically inputted during a predetermined test time period by time in the ground-real image generating step, It is preferable that the setting information of the optimization image analysis algorithm is applied to the image analysis of the field image inputted subsequently through the analysis optimization unit in consideration of the time.

Alternatively, in the ground-based image generation step, the ground-based actual image for the input scene image is generated for each environment according to a predetermined reference through the ground-real image generation unit, and at this time, Through the analysis optimization unit, the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each environment may be applied to the image analysis of the field image inputted afterwards.

Alternatively, in the ground-based image generation step, the ground-based actual image for the input scene image is generated for each season through the ground-real image generation unit, and at this time, in the image analysis optimization step, The setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each season may be applied to the image analysis of the field image inputted subsequently.

An apparatus and method for optimizing image analysis for a surveillance site according to an exemplary embodiment of the present invention receives a scene image taken from a surveillance camera, applies a plurality of image analysis algorithms to image analysis of a scene image, Is compared with the result of the ground truth image to generate an optimized image analysis algorithm. Thus, it is possible to easily construct an optimal monitoring environment for each monitoring site without the help of an expert.

An apparatus and method for optimizing image analysis for a surveillance site according to an exemplary embodiment of the present invention adjusts and tests parameter information and weight information for a plurality of image analysis algorithms, The image analysis algorithm for the monitoring site having the optimized parameters and the weighted values generated by the image analysis algorithm can be easily applied to the image analysis algorithm suitable for the individual monitoring sites.

The apparatus and method for optimizing an image analysis for a surveillance site according to an exemplary embodiment of the present invention is an image analysis algorithm optimized for each period and time through comparison with a ground image generated based on period, It is possible to apply the image analysis algorithm for the surveillance site that is optimized for the surveillance environment that changes according to the time zone of each surveillance site by applying the setting information of the surveillance site.

The apparatus and method for optimizing image analysis for a surveillance site according to an exemplary embodiment of the present invention can be applied to various monitoring environments such as early summer, late autumn, and seasonal or snow, rain, The image analysis algorithm for the surveillance site is optimized to improve the accuracy of the image analysis for the surveillance site.

The image analysis optimization apparatus and method for a surveillance site according to an exemplary embodiment of the present invention can be configured in various physical locations such as a surveillance camera, a DVR server, an NVR server, and a central control server, So that the efficiency of installation and maintenance can be improved.

1 is a block diagram of an image analysis optimizer for a surveillance site in accordance with an embodiment of the present invention.
FIG. 2, FIG. 3, and FIG. 4 illustrate application examples of an image analysis optimization apparatus for a surveillance site according to an embodiment of the present invention.
5 is a diagram illustrating an example of a comparison test cycle and a setting information application cycle of an image analysis optimization apparatus for a surveillance site according to an exemplary embodiment of the present invention.
FIG. 6 is an exemplary view showing matching of a plurality of image analysis algorithms with a ground image according to an embodiment of the present invention; FIG.
7 is a diagram illustrating an example of generation of an algorithm for optimizing image analysis through parameter information and weight information of a plurality of image analysis algorithms according to an embodiment of the present invention.
8 is a flowchart of an image analysis optimization method for a surveillance site according to an embodiment of the present invention.
9 is an illustration of a scene image of a surveillance site according to an embodiment of the present invention.
10 is an exemplary view of a ground-based image of a surveillance site according to an embodiment of the present invention;
11 is an exemplary view illustrating test results of a first image analysis algorithm applied to a field image of a surveillance site according to an embodiment of the present invention;
12 is an exemplary view of test results of a second image analysis algorithm applied to a field image of a surveillance site according to an embodiment of the present invention;
13 is an exemplary view of an image analysis result through an optimization image analysis algorithm according to an embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features and advantages of the invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.

1 is a block diagram of an image analysis optimization apparatus 100 for a surveillance site according to an embodiment of the present invention.

Referring to FIG. 1, an image analysis optimization apparatus 100 for a surveillance site according to an embodiment of the present invention includes a field image input unit 110 receiving a field image photographed from a surveillance camera, An image analysis algorithm management unit 130 for managing each setting information for a plurality of image analysis algorithms, a plurality of image analysis algorithms for managing the plurality of image analysis algorithms, An image analyzing algorithm testing unit 140 for adjusting and testing each of the setting information, and comparing the image analysis result of the test with the result of the ground image, A ground-based image matching unit 150 for generating an optimized image analysis algorithm having setting information, And an image analysis optimization unit 160 for applying the setting information of the optimization image analysis algorithm to the image analysis of the scene image.

In this case, the setting information preferably includes at least one of parameter information of each image analysis algorithm and weight information of each image analysis algorithm.

In addition, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention may include parameter information and weight information of the optimization image analysis algorithm for the scene image compared with the periodically generated image The optimization setting management unit 170 may classify and manage the setting information according to the period.

As a preferred embodiment, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention receives a field image photographed in the field from an SD class surveillance camera or an HD class surveillance camera according to a resolution classification, Or may receive a field image taken in the field from an infrared camera.

The ground truth image generated by the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention is excellent in accuracy of image analysis, It is difficult to directly apply it to the real-time field image.

Accordingly, as a preferred embodiment, the image analysis optimization apparatus 100 for a surveillance site according to an embodiment of the present invention includes a plurality of image analysis algorithms and a plurality of image analysis algorithms for real- (Including generation of setting information) optimized for a specific monitoring environment through a combination of a parameter change of an image analysis algorithm optimized through comparison with an image or a plurality of weighted image analysis algorithms, It is used for the image analysis of the field image of the field where the camera is installed.

Since the image analysis optimization apparatus 100 for a monitoring site according to an embodiment of the present invention automatically discriminates and applies the optimized image analysis algorithm through comparison with the ground image, , It is applicable in real time, but it has a low load, and the application and modification of the optimized image analysis algorithm are free, so that the image analysis is highly effective.

As a further preferred embodiment, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention includes a combination of fine adjustment of parameters of individual image analysis algorithms, weighting of each image analysis algorithm, And the optimization image analysis algorithm corresponding to the seasonal or environmental change is automatically changed so that the effectiveness in monitoring becomes very high.

As described above, the image analysis optimization apparatus 100 for a surveillance site according to an embodiment of the present invention receives a field image photographed from a surveillance camera and applies a plurality of image analysis algorithms to image analysis of a field image, The optimized image analysis algorithm is created by comparing the image analysis result of the image with the result of the ground truth image. Therefore, it is possible to easily construct an optimized surveillance environment for each monitoring site without the help of an expert.

FIGS. 2, 3 and 4 are diagrams illustrating application examples of an image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention. FIGS. 2 and 3 are views illustrating a surveillance site according to an exemplary embodiment of the present invention. FIG. 4 illustrates an example of the operation of the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention. Referring to FIG.

2, 3 and 4, an image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention physically includes a surveillance camera 30, a DVR server 50, The NVR server 60, and the central control server 200, and it is obvious that they may be physically separated from each other.

2, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention generally includes a DVR server 50 or a DVR server 50 The central control server 200 and the like can be configured.

3, if there is a configuration for image analysis in the IP camera 30, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention may be provided to the NVR server 60, The optimization image analysis algorithm may be determined and the configuration information such as the parameter information or the weight information may be transmitted to each individual IP camera 30 to apply an image analysis algorithm optimized for the individual monitoring sites have.

Alternatively, if it is determined that each individual surveillance camera 30 has sufficient load capacity, the individual surveillance camera 30 may be configured with the image analysis optimizing apparatus 100 for the surveillance site according to an embodiment of the present invention .

As described above, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention includes various physical locations such as a surveillance camera 30, a DVR server 50, an NVR server 60, and a central control server 200 And can be easily applied to a monitoring environment of a pre-established monitoring system, thereby improving the efficiency of installation and maintenance.

Referring to FIG. 4, an image analysis optimization apparatus 100 for a surveillance site according to an embodiment of the present invention receives a scene image from a surveillance camera 30 that monitors a specific site (surveillance region) And performs an optimization operation for image analysis according to a predetermined purpose, for example, an image analysis that tracks the movement of the person 1, for example.

As described above, the image analysis optimizing apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention can detect changes in illuminance (5), environmental changes (change in climate, weather, etc.) The optimization of the image analysis according to the change can be automatically performed.

The conventional video analysis configuration has a disadvantage that it is difficult to change after being set once and it is difficult to change the video environment due to noise, noise, weather, meaningful data such as shaking branches (2, 3) , The error rate of image analysis increased and the efficiency of surveillance dropped sharply.

In contrast, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention can detect a change in a specific environment (for example, a temporal change such as daytime and nighttime, It is very effective in monitoring by automatically generating the optimized image analysis algorithm in response to the seasonal change or by automatically applying the setting information. As the monitoring environment changes through temporal, environmental, and seasonal changes, So that the effectiveness of the image analysis can be maintained constant regardless of changes in the monitoring environment.

5 is a diagram illustrating an example of a comparison test period and a setting information application period of the image analysis optimization apparatus 100 for a surveillance site according to an embodiment of the present invention.

5, an image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention may periodically (7, 8, 9, 10, 11, 12) Can be performed.

Referring to FIGS. 1 and 5, in a preferred embodiment, the ground-based image generating unit 120 generates the ground-based real-time image of the field image periodically input for a predetermined test time, , The image analysis optimization unit 160 may apply the setting information of the optimization image analysis algorithm to the image analysis of the field image that is input after considering the time.

For example, as shown in FIG. 5A, after setting the test time to 24 hours (one day), the ground-based image generated repeatedly at predetermined intervals (7, 8, 9, 10, 11, It is possible to configure the image analysis to operate according to the setting information optimized for each time by checking the setting information of the optimization image analysis algorithm for each time.

Alternatively, as shown in FIG. 5B, after the test time is set to one month, the test time is repeatedly generated in a plurality of (for example, 7, 8, 9, 10) It is possible to configure the image analysis to correspond to the setting information optimized for each day or time after confirming the setting information of the optimized image analysis algorithm for each day or time by comparing the result values of the ground image and the image analysis algorithm have.

In this case, more preferably, the ground-side actual image matching unit 150 compares the comparison of the field image based on at least one of the image analysis result of the test and the result of the ground image, And may generate an optimized image analysis algorithm having the optimized setting information.

In another embodiment, the ground-based image generating unit 120 generates the ground-based actual image for the input scene image, for each season, and the image analysis optimizing unit 160 may generate the ground- And the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each season.

In another embodiment, the ground-based image generating unit 120 generates the ground-based actual image of the input scene image for each environment according to a preset reference, and the image analysis optimizing unit 160 , The setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each environment may be applied to the image analysis of the field image inputted afterwards.

For example, as shown in FIG. 5C, one or more ground-based actual images may be generated for each test period 21, 22, 23, and 24 classified by season or environment, and may be generated according to the season or environment corresponding to the generated ground- A suitable optimized image analysis algorithm may be generated through the above comparison (7, 8, 9, 10).

As a preferred embodiment, the ground-based image generating unit 120 may generate the ground-based real image for each environment according to a criterion including at least one of snow, rain, fog, and yellow dust.

As described above, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention tests image analysis algorithms by season, such as early summer, late fall, or an environment such as snow, rain, By applying the image analysis algorithm for the surveillance site optimized for the surveillance environment that changes according to the environment, the accuracy of the image analysis for the surveillance site is further improved.

FIG. 6 is an exemplary diagram illustrating matching of a plurality of image analysis algorithms with a ground image according to an exemplary embodiment of the present invention. In FIG. 6, a specific site (a surveillance site) according to time changes from 6:00 am to 6:00 pm FIG. 2 is a graph showing a matching degree of a plurality of image analysis algorithms.

6, an image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention includes a plurality of image analysis algorithms 71, 72, 73, and 74 72, 73, and 74, and determines the accuracy of each of the image analysis algorithms 71, 72, 73, and 74, Or by adjusting the parameter information, it is possible to generate an image analysis algorithm optimized for each time.

As a preferred embodiment, since the degree of matching of the algorithm A 71 is high in the vicinity of 6:00 am in FIG. 6, the first optimization algorithm generated by adjusting the parameter values of the algorithm A 71 The algorithm B (72) and the algorithm D (74) have a higher degree of matching than the algorithm A (71) or the algorithm C (73) from 11:00 am to 1:00 am B 72 and Algorithm D 74 according to a predetermined weighting criterion to generate and apply a second optimization algorithm.

As another preferred embodiment, since the degree of matching of the algorithm C (73) and the algorithm D (74) is higher than that of the algorithm A (71) or the algorithm B (72) ) And the algorithm D 74, and then generates and applies a third optimization algorithm generated by appropriately combining the weights of the respective algorithms 73 and 74.

It should be noted that, as described above, the adjustment of the optimization parameters and the adjustment of the weights can be automatically applied not only by time, but also by season and environment, through the comparison of the comparison with the ground image.

As described above, the image analysis optimization apparatus 100 for a surveillance site according to an exemplary embodiment of the present invention is optimized for each period and time through comparison with a ground-based image generated based on period, time, or cumulative information By applying the configuration information of the image analysis algorithm, it is possible to apply the image analysis algorithm for the surveillance site which is optimized for the surveillance environment that changes according to the time of each surveillance site, thereby improving the efficiency of constructing surveillance environment for the surveillance site.

7 is an exemplary diagram illustrating generation of an optimized image analysis algorithm based on parameter information and weight information of a plurality of image analysis algorithms according to an embodiment of the present invention.

7, an image analysis optimization apparatus 100 for a surveillance site according to another embodiment of the present invention receives a scene image taken from a surveillance camera, adjusts at least one parameter information and weight information, The image analyzing algorithms 15, 16, 17 and 18 are applied to the image analysis of the field image, and the image analysis results of the test are compared with the results of the ground truth image of the field image Generating an optimized image analysis algorithm (19) having the optimized parameter information and weight information, and applying the optimized image analysis algorithm having the parameter information and weight information optimized for image analysis on the input field image 20).

In a preferred embodiment, as shown in the example, the parameter information may be information that is numerically adjusted or information that is adjusted to attributes such as selection, release, or information that is adjusted to attributes other than numbers, such as yes / .

Also, as a preferred embodiment, the weight information may be information indicating the ratio or degree of reflection of each image analysis algorithm (15, 16, 17, 18) tested for the optimized image analysis algorithm to be generated.

As described above, the image analysis optimization apparatus 100 for a surveillance site according to an embodiment of the present invention adjusts and tests parameter information and weight information for a plurality of image analysis algorithms, An image analysis algorithm for a surveillance site having optimized parameters and weights composed of a plurality of image analysis algorithms is generated through the image analysis algorithm, so that an image analysis algorithm suitable for an individual surveillance site can be easily and automatically applied.

8 is a flowchart illustrating an image analysis optimization method for a surveillance site according to an exemplary embodiment of the present invention.

Referring to FIG. 8, an image analysis optimization method for a surveillance site according to an embodiment of the present invention manages respective setting information for a plurality of image analysis algorithms through an image analysis algorithm management unit 130.

In addition, a field image photographed from a surveillance camera is inputted through a field image input unit 110 (S10).

Then, a ground truth image for the scene image is generated through the ground image generating unit 120 (S20).

At this time, the ground-based image generator 120 can generate the ground-based real-time image of the field image periodically input for a predetermined test time, on a time-by-time basis.

Alternatively, the ground-based actual image of the field image input through the ground-real-image generating unit 120 may be generated for each environment according to a preset reference.

Alternatively, the ground-based actual image for the field image input through the ground-real-image generating unit 120 may be generated for each season.

Thereafter, the image analysis algorithm testing unit 140 applies the plurality of image analysis algorithms to the image analysis of the field image, and adjusts and tests each of the setting information (S30, S40).

Thereafter, the image analysis result of the test is compared with the result value of the ground image (S50) with respect to the field image through the ground image matching unit 150 to generate an optimized image analysis algorithm having the setting information optimized (S60).

Then, the setting information of the optimization image analysis algorithm is applied to the image analysis of the field image, which is input afterwards through the image analysis optimization unit 160 (S70).

In this case, it is preferable that the setting information of the optimization image analysis algorithm is applied to the image analysis of the field image inputted after the image analysis optimization unit 160 considering the time.

Alternatively, the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each environment may be applied to the image analysis for the field image, which is input after the image analysis optimization unit 160 You may.

In addition, the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each season is applied to the image analysis for the field image, which is input through the image analysis optimization unit 160 It is possible.

Meanwhile, the image analysis optimization method for a surveillance site according to an exemplary embodiment of the present invention may include a parameter of the optimization image analysis algorithm for the field image, which is periodically generated through the optimization setting management unit 170 and compared with the ground image Information and weight information as the setting information, and classifies and manages the information according to the period.

9 to 13 illustrate examples in which the image analysis optimization method according to an embodiment of the present invention is applied. At this time, it is assumed that the optimization of the image analysis proceeds to optimize the image analysis for recognizing a person in a field image.

FIG. 9 is a view illustrating an example of a field image of a surveillance site according to an embodiment of the present invention.

10 is an example of a ground-based image of a surveillance site according to an embodiment of the present invention, and is an example of generating a ground-based image of the site image.

As shown in FIG. 10, the generated ground image is excellent in the accuracy of the image analysis, but consumes a lot of time and resources to generate it, and thus can not be applied directly to the real-time scene image.

11 is an example of test results of a first image analysis algorithm applied to a field image of a surveillance site according to an embodiment of the present invention.

As a preferred embodiment, FIG. 11 shows an example of performing an image analysis test for human recognition by applying a background difference algorithm as the first image analysis algorithm.

12 is an example of test results of a second image analysis algorithm applied to a field image of a surveillance site according to an embodiment of the present invention.

As a preferred embodiment, FIG. 12 shows an example in which an image analysis test for human recognition is performed by applying a Gaussian Mixture Algorithm (Gaussian Mixture Algorithm) as the second image analysis algorithm.

11 and 12, in the field image of the surveillance site, the test result of the Gaussian mixture model algorithm, which is the second image analysis algorithm, is compared with the background difference algorithm which is the first image analysis algorithm at the specific time Is relatively high.

As a preferred embodiment, in the image analysis optimization method according to an embodiment of the present invention, the result of the ground image of FIG. 10 and the test image of the first image analysis algorithm and the second image analysis algorithm performed in FIGS. And adjusting the setting information of each image analysis algorithm on the basis of the matching degree according to the preset reference, thereby generating an image analysis algorithm optimized for the monitoring site.

FIG. 13 is a diagram illustrating an image analysis result through an optimization image analysis algorithm according to an embodiment of the present invention.

As shown in FIG. 13, the image analysis optimization method according to an embodiment of the present invention combines the first image analysis algorithm and the second image analysis algorithm, and adjusts and optimizes parameter information of each image analysis algorithm The optimization image analysis algorithm can be generated.

As a preferred embodiment, the image analysis algorithm for deriving the image analysis result shown in FIG. 13 is based on Chroma Noise Removal, Showdow Removal, and Morphology of the Gaussian mixture model algorithm, which is the second image analysis algorithm Mophplogy) may be blended to remove the illumination, color noise, and shadow.

As described above, more preferably, when the specific parameter information of the first image analysis algorithm is changed and the degree of matching through combination with the second image analysis algorithm is higher, the first image analysis algorithm And the second image analysis algorithm may be appropriately combined to generate and apply an optimized image analysis algorithm.

The foregoing embodiments and advantages are merely exemplary and are not to be construed as limiting the present invention. However, the present invention is not limited to the above-described embodiments, and various changes and modifications may be made by those skilled in the art without departing from the scope of the present invention. .

1: person 2: tree
3: Leaves 4: Rain
5: Street light 30: Field surveillance camera
50: DVR server 60: NVR server
100: Image analysis optimization apparatus 110: Field image input unit
120: ground image generation unit 130: image analysis algorithm management unit
140: image analysis algorithm testing unit 150: ground image matching unit
160: Image analysis optimization unit 170: Optimization setting management unit
200: Central control server

Claims (15)

A field image taken from a surveillance camera, adjusting at least one parameter information and weight information, and applying a plurality of image analysis algorithms to image analysis of the field image,
Comparing an image analysis result of the test with a result of a ground truth image of the field image to generate an optimized image analysis algorithm having the parameter information and the weight information optimized,
And applying the optimized image analysis algorithm having the parameter information and the weight information optimized for the image analysis of the field image, which is input afterward, to the monitoring site.
A field image input unit for receiving a field image photographed from a surveillance camera;
A ground image generating unit for generating a ground truth image of the field image;
An image analysis algorithm management unit for managing respective setting information for a plurality of image analysis algorithms;
An image analysis algorithm testing unit applying the plurality of image analysis algorithms to image analysis of the field image, and adjusting and testing each of the setting information;
A grounded image matching unit for generating an optimized image analysis algorithm having the setting information optimized by comparing the image analysis result of the test with the result of the grounded actual image with respect to the field image;
And an image analysis optimization unit for applying the setting information of the optimization image analysis algorithm to an image analysis of the field image inputted afterwards.
3. The method of claim 2,
The parameter information of each image analysis algorithm, and the weight information of each image analysis algorithm.
3. The method of claim 2,
Wherein the ground-based image generating unit generates the ground-based real-time image for the field image periodically input for a predetermined test time,
Wherein the image analysis optimizing unit applies the setting information of the optimization image analysis algorithm to the image analysis of the field image inputted afterwards, in consideration of the time.
5. The image processing apparatus according to claim 4, wherein the ground-
The optimization image analysis algorithm having the setting information optimized by comparing and comparing based on at least one of the image analysis result of the test and the result of the ground image is generated in the comparison of the field image Image analysis optimization device for monitoring sites.
3. The method of claim 2,
The ground-based image generating unit generates the ground-based actual image of the input scene image for each environment according to a preset reference,
Wherein the image analysis optimizing unit applies the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each environment to the image analysis for the scene image inputted subsequently Image analysis optimization device for the site.
7. The image processing apparatus according to claim 6, wherein the ground-
Wherein the ground image is generated for each environment according to a criterion including at least one of snow, rain, fog, and yellow dust.
3. The method of claim 2,
Wherein the ground-based image generating unit generates the ground-based actual image of the input scene image for each season,
Wherein the image analysis optimizing unit applies the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each season to the image analysis for the scene image inputted subsequently. Image analysis optimization device for.
3. The method of claim 2,
And an optimization setting management unit for classifying and managing parameter information and weight information of the optimized image analysis algorithm for the scene image compared with the periodically generated ground image, Image analysis optimization device for.
The apparatus of claim 2, wherein the image analysis optimization apparatus for the monitoring site
Characterized in that it is physically located at one of a surveillance camera, a DVR server, an NVR server, and a central control server.
a) receiving a scene image taken from a surveillance camera through a scene image input unit;
b) generating a ground truth image for the scene image through a ground image generating unit;
c) managing each setting information for a plurality of image analysis algorithms through an image analysis algorithm management unit;
d) applying, through an image analysis algorithm testing unit, the plurality of image analysis algorithms to the image analysis of the field image, adjusting and testing each of the setting information;
e) generating, for the field image, an optimized image analysis algorithm having the setting information optimized by comparing the image analysis result of the test with the result of the ground image through the ground image matching unit;
and f) applying the setting information of the optimized image analysis algorithm to the image analysis of the field image, which is input afterwards, through the image analysis optimization unit.
12. The method of claim 11,
g) classifying and managing, as the setting information, the parameter information and the weight information of the optimized image analysis algorithm for the scene image compared with the periodically generated ground image through the optimization setting management unit; The method comprising the steps of:
12. The method of claim 11,
Wherein the step b) comprises: generating the ground-based actual image of the field image periodically inputted for a predetermined test time on a time basis through the ground-real image generating unit,
Wherein, in the step (f), the setting information of the optimization image analysis algorithm is applied to the image analysis of the field image inputted after the image analysis optimization unit, considering the time. Optimization method.
12. The method of claim 11,
Wherein the step b) comprises: generating the ground-based actual image of the input scene image by the ground-based real image generating unit for each environment according to a preset reference,
Wherein the step (f) comprises the step of: applying the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each environment A method for optimizing image analysis for a surveillance site.
12. The method of claim 11,
Wherein the step b) comprises: generating the ground-based actual image of the field image input by the seasonally-generated image generating unit,
Wherein the step (f) comprises the step of applying the setting information of the optimization image analysis algorithm generated through the comparison of the test result values generated for each season to the image analysis for the scene image input through the image analysis optimization unit A method for image analysis optimization for a monitoring site.
KR1020130067279A 2013-06-12 2013-06-12 Video analysis optimization apparatus and method for surveillance site KR20140144977A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
KR1020130067279A KR20140144977A (en) 2013-06-12 2013-06-12 Video analysis optimization apparatus and method for surveillance site

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
KR1020130067279A KR20140144977A (en) 2013-06-12 2013-06-12 Video analysis optimization apparatus and method for surveillance site

Publications (1)

Publication Number Publication Date
KR20140144977A true KR20140144977A (en) 2014-12-22

Family

ID=52674996

Family Applications (1)

Application Number Title Priority Date Filing Date
KR1020130067279A KR20140144977A (en) 2013-06-12 2013-06-12 Video analysis optimization apparatus and method for surveillance site

Country Status (1)

Country Link
KR (1) KR20140144977A (en)

Similar Documents

Publication Publication Date Title
US8890959B2 (en) Configuration of image capturing settings
US20160260306A1 (en) Method and device for automated early detection of forest fires by means of optical detection of smoke clouds
US11394863B2 (en) No-reference image quality assessment for iterative batch video analysis
DE112013005195T5 (en) Method and apparatus for selecting a video analysis algorithm based on historical event data
TW201904265A (en) Abnormal motion detection method and system
CN105208325A (en) Territorial resource monitoring and early warning method based on image fixed-point snapshot and comparative analysis
US20130128080A1 (en) Image analysis device and method thereof
Nagothu et al. Authenticating video feeds using electric network frequency estimation at the edge
GB2552511A (en) Dynamic parametrization of video content analytics systems
KR20160089165A (en) System and Method for Detecting Moving Objects
Kong et al. A new quality model for object detection using compressed videos
CN106781167A (en) The method and apparatus of monitoring object motion state
US10593047B2 (en) Setting of a motion trigger level
CN106682590A (en) Processing method and server for monitoring service
KR20140144977A (en) Video analysis optimization apparatus and method for surveillance site
CN116193251A (en) Self-adaptive security camera monitoring method and system
CN108692709A (en) A kind of farmland the condition of a disaster detection method, system, unmanned plane and cloud server
US11153474B2 (en) Automated scope limiting for video analytics
JP2016134804A (en) Imaging range abnormality discrimination device, imaging range abnormality discrimination method and computer program for imaging range abnormality discrimination
Liang et al. Real time intrusion detection system for outdoor environment
KR101237524B1 (en) Output expression system for multi-input of facility water management system
EP4052460A1 (en) No-reference image quality assessment for iterative batch video analysis
Kim A Study on Edge Board for Blur Image Judgment and Cluster Image Behavior Analysis: AI-Based Image Processing System Research
Khan et al. MobiTrick–Mobile Traffic Checker
GB2521267A (en) A method of assessing sensor performance

Legal Events

Date Code Title Description
A201 Request for examination
E902 Notification of reason for refusal
E601 Decision to refuse application