KR20240040005A - Apparatus and method for analyzing image - Google Patents

Abstract

The present invention relates to an image analysis device and method, which calculates the absolute distance of a reference map object among map objects included in a depth map, applies the absolute distance to the remaining portion of the depth map, and It relates to an image analysis device and method for calculating the absolute distance for the whole.
An image analysis device according to an embodiment of the present invention includes an imaging unit that generates a captured image, an object detector that detects an object in the captured image, and a depth map generator that generates a depth map for the captured image. , and select a reference object among the detected objects, and refer to the absolute distance calculated for the reference map object corresponding to the reference object among the map objects included in the depth map, and the absolute distance for the entire area of the depth map. It includes a map analysis unit that calculates.

Description

Image analysis device and method {Apparatus and method for analyzing image}

The present invention relates to an image analysis device and method, and more specifically, to calculating the absolute distance of a reference map object among map objects included in a depth map and applying the absolute distance to the remaining portion of the depth map. , relates to an image analysis device and method for calculating the absolute distance for the entire depth map.

A depth map represents information including the distance between an observation point and an object. Because the depth map contains distance information between the observation point and each surface of the object, it is possible to calculate a 3D image of the object using the depth map.

Meanwhile, since the depth map contains only relative distance information between the observation point and the object, the absolute distance of the object included in the depth map cannot be calculated through this.

Therefore, there is a need for an invention that allows calculating the absolute distance of an object included in a depth map.

U.S. Patent Publication No. 2011-0142287 (2011.06.16)

The problem to be solved by the present invention is to calculate the absolute distance of the reference map object among the map objects included in the depth map, and apply the absolute distance to the remaining part of the depth map to determine the absolute distance for the entire depth map. To provide an image analysis device and method for calculating distance.

The objects of the present invention are not limited to the problems mentioned above, and other objects not mentioned will be clearly understood by those skilled in the art from the description below.

In order to achieve the above problem, an image analysis device according to an embodiment of the present invention includes an imaging unit that generates a captured image, an object detector that detects an object in the captured image, and a depth map for the captured image. A depth map generator that generates, and selects a reference object from among the detected objects, and maps the depth map with reference to an absolute distance calculated with respect to a reference map object corresponding to the reference object among map objects included in the depth map. It includes a map analysis unit that calculates the absolute distance for the entire area.

The depth map includes relative distance information of objects included in the captured image.

The object detection unit detects a reference object for which standard size information is secured among objects included in the captured image, and the map analysis unit selects the reference object from among the reference objects.

The map analysis unit includes a distance calculation unit that calculates an absolute distance of the reference map object with reference to the size of the reference map object, and an entire depth map according to the ratio of the relative distance and the absolute distance of at least two reference map objects. It includes a distance conversion unit that converts the relative distance to the area into an absolute distance.

The distance calculation unit calculates the absolute distance of the reference map object by referring to the size of a reference area included in a preset part of the reference map object.

The distance calculation unit extracts some of the relative distance information included in the reference area and calculates the relative distance of the reference map object.

The distance calculation unit creates a feature area of a preset size in the reference area, and determines the average or median value of the relative distances included in the feature area as the relative distance of the reference map object.

The map analysis unit detects a reference object in a captured image for a preset area of interest among the captureable areas of the imaging unit, and determines a pre-generated absolute distance with respect to the depth map of the captured image based on the detected reference object. Update .

The captureable area includes a plurality of areas of interest, and the map analysis unit detects a reference object from a captured image of an area of interest switched according to a preset order among the plurality of areas of interest.

The object detection unit detects an object in the captured image in real time, and when an object is detected in the captured image according to the real-time detection result, the map analysis unit calculates an absolute distance for the entire area of the depth map.

An image analysis method according to an embodiment of the present invention includes generating a captured image, detecting an object in the captured image, generating a depth map for the captured image, and the detected object. selecting a reference object among objects, and calculating an absolute distance for the entire area of the depth map by referring to an absolute distance calculated for a reference map object corresponding to the reference object among map objects included in the depth map. Includes steps.

The depth map includes relative distance information of objects included in the captured image.

The step of detecting the object includes detecting a reference object for which standard size information is secured among objects included in the captured image, and the step of selecting the reference object includes selecting the reference object from among the reference objects. Includes.

Calculating the absolute distance for the entire area of the depth map includes calculating the absolute distance of the reference map object with reference to the size of the reference map object, and the relative distance and absolute distance of at least two reference map objects. Converting the relative distance for the entire area of the depth map to an absolute distance according to the ratio of

Calculating the absolute distance of the reference map object includes calculating the absolute distance of the reference map object with reference to the size of a reference area included in a preset portion of the reference map object.

The image analysis method further includes calculating the relative distance of the reference map object by extracting some of the relative distance information included in the reference area.

The step of calculating the relative distance of the reference map object includes creating a feature area of a preset size in the reference area, and calculating the average or median value of the relative distance included in the feature area to the relative distance of the reference map object. It includes determining what the distance is.

The image analysis method includes detecting a reference object in a captured image for a preset region of interest among the captureable areas of the imaging unit that generates the captured image, and creating a depth map of the captured image based on the detected reference object. It further includes updating the absolute distance created in advance.

The captured image includes a plurality of regions of interest, and the step of detecting a reference object in the region of interest includes detecting a reference object in the captured image of the region of interest switched according to a preset order among the plurality of regions of interest. Includes.

The step of detecting an object in the captured image includes detecting an object in the captured image in real time, and the step of calculating an absolute distance for the entire area of the depth map includes the step of calculating an absolute distance in the captured image according to the real-time detection result. and calculating an absolute distance for the entire area of the depth map when an object is detected.

Specific details of other embodiments are included in the detailed description and drawings.

According to the image analysis device and method of the present invention as described above, the absolute distance of a reference map object for which a correlation between size and absolute distance is formed among map objects included in a depth map is calculated, and the absolute distance is calculated. It has the advantage of being able to calculate the absolute distance for the entire depth map by applying to the remaining part of the depth map.

In addition, since it is implemented only in software without separate operation by hardware, it has the advantage that the process of calculating the absolute distance for the entire depth map is performed relatively quickly.

The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

1 is a block diagram of an image analysis device according to an embodiment of the present invention.
Figure 2 is a block diagram of the map analysis unit.
Figure 3 is a diagram showing a captured image.
Figure 4 is a diagram showing a reference object being detected in a captured image.
Figure 5 is a diagram showing a depth map of a captured image.
Figure 6 is a diagram showing a reference box being formed in a portion of a reference map object.
Figure 7 is a diagram showing a reference area being formed in a reference box.
Figure 8 is a diagram showing a reference box being formed throughout the reference map object.
Figure 9 is a diagram showing a reference point displayed on a distance coordinate plane.
Figure 10 is a diagram showing a reference graph being formed based on reference points.
Figure 11 is a diagram showing a reference graph being formed based on three reference points.
Figure 12 is a flowchart showing an image analysis method according to an embodiment of the present invention.
Figure 13 is a diagram showing the area of interest being switched.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. The advantages and features of the present invention and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. The present embodiments are merely intended to ensure that the disclosure of the present invention is complete and to provide common knowledge in the technical field to which the present invention pertains. It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings that can be commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless clearly specifically defined.

1 is a block diagram of an image analysis device according to an embodiment of the present invention.

Referring to FIG. 1, the image analysis device 100 according to an embodiment of the present invention includes an imaging unit 110, a storage unit 120, a control unit 130, an object detection unit 140, and a depth map generator 150. , and includes a map analysis unit 160 and an output unit 170.

The imaging unit 110 can generate captured images. For example, the imaging unit 110 may be a camera equipped with an image sensor such as a Complementary Metal-Oxide-Semiconductor (CMOS) or Charge-Coupled Device (CCD).

In the present invention, the imaging unit 110 may be a monocular camera. The captured image generated by the imaging unit 110 indicates that it was captured at one shooting point.

Additionally, the imaging unit 110 may be provided with a lens unit (not shown) capable of auto focus. As will be described later, the imaging unit 110 may perform autofocus under the control of the map analysis unit 150.

Additionally, the imaging unit 110 may be capable of adjusting the shooting direction. For example, the imaging unit 110 may include pan tilt equipment (not shown). The pan tilt equipment can adjust the shooting direction of the imaging unit 110 based on an input control signal.

The storage unit 120 can temporarily or permanently store the captured image generated by the imaging unit 110. Additionally, the storage unit 120 may store the depth map generated by the depth map generator 150 and the analysis result generated by the map analysis unit 160.

Additionally, the storage unit 120 may include a table (hereinafter referred to as an absolute distance table) (not shown) containing the correlation between the size of the object and the absolute distance. In general, people, bicycles, vehicles, etc. can be formed to have a constant size. In this way, when an object whose size is generally uniform is included in a captured image, the absolute distance of the depth map object to the object may be determined using the size of the object. The storage unit 120 may store an absolute distance table including the relationship between the size of the object and the absolute distance, which may be used by a distance calculation unit to be described later.

The object detection unit 140 functions to detect objects in captured images. For this purpose, the object detection unit 140 may use an object detection model. For example, the object detector 140 may detect an object in a captured image using the YOLO (You Only Live Once) algorithm as an object detection model. The YOLO algorithm is an artificial intelligence algorithm suitable for surveillance cameras that process real-time video because the speed of detecting objects in captured video is relatively fast. The results output by the YOLO algorithm may include a bounding box indicating the location of each object and a classification probability of what the object is.

The depth map generator 150 serves to generate a depth map for the captured image generated by the imaging unit 110. For example, the depth map generator 150 may generate a depth map of a captured image generated by the imaging unit 110, which is a monocular camera, using a deep learning-based model.

The depth map may include relative distance information of objects included in the captured image. The depth map may include not only relative distance information between objects included in the image, but also relative distance information across surfaces constituting each object. For example, a depth map can be understood as containing relative distance information for all pixels included in a captured image. By using the relative distance information of objects, it is possible to estimate not only the arrangement relationship between objects but also the three-dimensional shape of each object.

Alternatively, according to some embodiments of the present invention, the depth map generator 140 generates a depth map using an artificial intelligence model and then reflects the characteristics of the focus lens (not shown) provided in the imaging unit 110. You can transform the depth map. To explain this in detail, the absolute distance between the object photographed by the imaging unit 110 and the imaging unit 110 is calculated by referring to the position of the focusing lens, and the depth map generator 140 determines the location and absolute distance of the focusing lens. The depth map can be transformed to reflect the amount of change between distances.

Additionally, the depth map generator 140 may generate a depth map using at least one artificial intelligence model selected from a plurality of different artificial intelligence models. Multiple artificial intelligence models can generate optimal depth maps in different situations. For example, the first artificial intelligence model can generate an optimal depth map for an image containing a daytime scene, and the second artificial intelligence model can generate an optimal depth map for an image containing a nighttime scene. there is. Alternatively, the third artificial intelligence model may generate an optimal depth map for images containing people or animals, and the third artificial intelligence model may generate an optimal depth map for images containing a natural environment. The depth map generator 140 may select at least one of a plurality of artificial intelligence models with reference to the shooting conditions by the imaging unit 110 and generate a depth map using the selected artificial intelligence model.

The map analysis unit 160 analyzes the depth map and calculates the absolute distance for the entire area of the depth map. Specifically, the map analysis unit 160 selects a reference object among the objects detected by the object detection unit 140, and calculates the absolute distance to the reference map object corresponding to the reference object among the map objects included in the depth map. By referring to it, you can calculate the absolute distance for the entire area of the depth map. The detailed configuration and function of the map analysis unit 160 will be described later with reference to FIGS. 2 to 8.

The output unit 170 may output the analysis result of the map analysis unit 160. The user can check the absolute distance for the entire area of the captured image by referring to the analysis result of the map analysis unit 160. At this time, the object detection unit 140 may detect the object based on the captured image generated without direction change or enlargement by the imaging unit 110, and the depth map generator 150 may generate a depth map. Ultimately, according to the present invention, it is possible to output analysis results by the map analysis unit 160 without switching the captured image being displayed.

The control unit 130 performs overall control over the imaging unit 110, storage unit 120, object detection unit 140, depth map creation unit 150, map analysis unit 160, and output unit 170. . For example, when a user command for calculating an absolute distance is input, the control unit 130 controls the imaging unit 110, the object detection unit 140, the depth map generation unit 150, and the map analysis unit 160. The absolute distance of the depth map for the captured image can be generated.

Meanwhile, according to some embodiments of the present invention, the control unit 130 may generate the absolute distance of the depth map for the captured image without inputting a user command. In this case, the object detection unit 140 can detect the object in real time from the captured image. Thus, when an object is detected in the captured image according to the real-time detection result of the object detection unit 140, the map analysis unit 160 can calculate the absolute distance for the entire area of the depth map. When a new absolute distance is calculated, the map analysis unit 160 may update the previously generated absolute distance with the new absolute distance, and this process may be performed repeatedly.

Figure 2 is a block diagram of the map analysis unit.

Referring to FIG. 2, the map analysis unit 160 includes a distance calculation unit 161 and a distance conversion unit 162.

The distance calculation unit 161 serves to calculate the absolute distance of the reference map object by referring to the size of the reference map object. As described above, an absolute distance table may be stored in the storage unit 120. The distance calculation unit 161 may calculate the absolute distance of the reference map object by referring to the absolute distance table. The distance calculation unit 161 may calculate the absolute distance of the reference map object by extracting the absolute distance corresponding to the size of the reference map object from the absolute distance table.

In the present invention, the reference object represents an object used to convert the relative distance for the entire area of the depth map into an absolute distance. The reference object may be randomly selected from objects included in the captured image.

The distance converter 162 serves to convert the relative distance for the entire area of the depth map into an absolute distance according to the ratio of the relative distance and absolute distance of at least two reference map objects. If the relative distance and absolute distance for two or more reference map objects are confirmed, the ratio between the relative distance and absolute distance for the entire depth map can be calculated through this. The distance converter 162 may calculate the absolute distance of each area by applying the relative distance of the entire area constituting the depth map to the ratio between the relative distance and the absolute distance.

FIG. 3 is a diagram showing a captured image, and FIG. 4 is a diagram showing a reference object being detected in a captured image.

Referring to FIG. 3, the imaging unit 110 may generate a captured image 200 by photographing the front.

The captured image 200 may include at least one object 210. Objects 210 may include people, animals, vehicles, or buildings. The captured image 200 is generated by the imaging unit 110 and may be a still image or one of a plurality of scenes included in a video.

Referring to FIG. 4 , the object detector 140 may detect an object 210 included in the captured image 200.

In particular, the object detection unit 140 of the image analysis device 100 according to an embodiment of the present invention can detect a reference object. In the present invention, the reference object represents an object for which standard size information is secured among the objects 210 included in the captured image 200. People, bicycles, vehicles, etc. can be formed to have uniform sizes. In this way, an object whose size is generally uniform may be a reference object, and the object detector 140 may detect such a reference object in the captured image 200. When a reference object is detected, a bounding box 220 may be formed at the edge of the reference object. The map analysis unit 160 described above may select a reference object from reference objects.

FIG. 5 is a diagram showing a depth map of a captured image, FIG. 6 is a diagram showing a reference box being formed in a part of a reference map object, FIG. 7 is a diagram showing a reference area being formed in a reference box, and FIG. 8 is a diagram showing that a reference box is formed throughout the reference map object.

Referring to FIG. 5 , the depth map generator 150 may generate a depth map 300 for the captured image 200.

The depth map 300 may include relative distance information of the object 310. Relative distance information represents the relative distance between each point included in the depth map 300 and the shooting point.

In the present invention, the resolution of the depth map 300 that can be processed by the map analysis unit 150 may be different from the resolution of the captured image 200. In this case, the depth map generator 140 may convert the resolution of the captured image 200 to correspond to the resolution of the preset depth map 300 and generate the depth map 300 using the converted captured image. there is.

Referring to FIGS. 5 and 6 , the distance calculation unit 161 may extract a reference map object 310 corresponding to the reference object and create a reference area 410 in the reference map object 310 .

The distance calculation unit 161 may calculate the absolute distance of the reference map object 310 by referring to the size of the reference area 410 included in the preset portion 311 of the reference map object 310. For example, when the reference object is a person, the distance calculation unit 161 creates a reference area 410 in a part of the reference map object 310 (hereinafter referred to as the target object) 311 corresponding to the person's head. You can.

The reference area 410 may be provided in a rectangular shape with the maximum size included in the target object 311. For example, four corners of the reference area 410 may overlap the boundary of the target object 311.

As the reference area 410 is provided in a rectangular shape, the size of the reference area 410 can be calculated using the horizontal length (H1) and vertical length (V1) of the reference area 410. The distance calculation unit 161 may calculate the absolute distance of the reference map object by referring to the size of the reference area 410.

Referring to FIG. 7 , the distance calculation unit 161 may calculate the relative distance of the reference map object 310 by extracting some of the relative distance information included in the reference area 410.

The distance calculation unit 161 may create a feature area 411 inside the reference area 410. The distance calculation unit 161 may generate the feature area 411 with a preset size. For example, the distance calculation unit 161 calculates the horizontal length (H2) and vertical length ( V2) can be set.

The distance calculation unit 161 may create a feature area 411 at the center of the reference area 410. For example, the distance calculation unit 161 may generate the feature area 411 so that the center C of the reference area 410 and the center of the feature area 411 coincide. Meanwhile, according to some embodiments of the present invention, the distance calculation unit 161 may generate the feature area 411 so that the feature area 411 is placed at an arbitrary point of the reference area 410.

The distance calculation unit 161 may calculate the average or median value of the relative distances included in the feature area 411 and determine this to be the relative distance of the reference map object 310. Here, the average value represents the sum of all relative distances included in the feature area 411 divided by the number of relative distances, and the median value represents the value corresponding to the middle of the distribution of relative distances included in the feature area 411. . Alternatively, the median value may represent the relative distance of the center pixel of the feature area 411. However, in the present invention, the relative distance of the reference map object 310 is determined as the average or median value of the relative distances included in the feature area 411, and various values representing the feature area 411 are used as the standard. The map object 310 may be determined as a relative distance. Hereinafter, the description will focus on how the relative distance of the reference map object 310 is determined by the average or median value of the feature area 411.

The absolute distance and relative distance of the reference map object 310 are calculated by the distance calculation unit 161, and the distance conversion unit 162 converts the relative distance for the entire area of the depth map into an absolute distance based on this. You can.

Meanwhile, FIGS. 6 and 7 illustrate that the reference area 410 is created in the target object 311, which is part of the reference map object 310, but as shown in FIG. 8, at the edge of the reference map object 310 A reference area 420 may be created. In this case, the reference area 420 may be created with the same size as the bounding box 220 of the reference object, and the distance converter 162 may calculate the absolute distance and relative distance of the reference map object 310 based on this. You can.

FIG. 9 is a diagram showing a reference point displayed on a distance coordinate plane, and FIG. 10 is a diagram showing a reference graph being formed based on the reference point.

Referring to FIG. 9, the distance calculation unit 161 may display the reference points P1 and P2 on the distance coordinate plane.

The distance coordinate plane can be formed by a horizontal axis representing absolute distance and a vertical axis representing relative distance. The reference points (P1, P2) represent coordinate points on the distance coordinate plane corresponding to the reference map object 310.

The distance calculation unit 161 may select a reference object from reference objects. At least two reference objects may be selected from different reference objects.

Additionally, the distance calculation unit 161 may select reference objects such that the relative distance between different reference map objects 310 is greater than or equal to a preset interval. FIG. 9 shows that the relative distance interval between two reference map objects 310 is D.

When a reference object is selected, the distance calculation unit 161 creates a reference area 410 in the reference map object 310 corresponding to the reference object, and refers to the size of the reference area 410 to create the reference map object 310. The absolute distance can be calculated. In addition, the distance calculation unit 161 may form a feature area 411 in the reference area 410 and calculate the relative distance of the reference map object 310 by referring to the relative distance included in the feature area 411. there is. At this time, if the interval between the relative distances of different reference map objects 310 is less than the preset interval, the distance calculation unit 161 selects a new reference object and repeats the above process based on the new reference object. You can.

When the absolute distance and relative distance of the reference map object 310 are confirmed, reference points P1 and P2 corresponding to the reference object may be displayed on the distance coordinate plane. Figure 9 shows two reference points (P1, P2) for two reference objects displayed on the distance coordinate plane. Since the relative distance and absolute distance of the first reference map object are A1 and B1, respectively, the first reference point P1 may be displayed on the distance coordinate plane based on this. Likewise, since the relative distance and absolute distance of the second reference map object are A2 and B2, respectively, the second reference point P2 may be displayed on the distance coordinate plane based on this.

Referring to FIG. 10 , the distance converter 162 may generate a reference graph G1 based on the reference points P1 and P2.

The reference graph (G1) can be created by connecting a plurality of reference points (P1, P2). Figure 10 shows that two reference points (P1, P2) are connected to create a straight reference graph (G1).

The reference graph G1 reflects the ratio of the absolute distance and the relative distance of the reference map object 310, and the distance converter 162 can use this to calculate the absolute distance to another part of the depth map. 10, if the relative distance is a1, the absolute distance of the corresponding point in the depth map is determined as b1, if the relative distance is a2, the absolute distance of the corresponding point in the depth map is determined as b2, and the relative distance is a3 In this case, the absolute distance of the corresponding point in the depth map can be determined as b3.

The distance converter 162 can calculate the absolute distance of the entire area 400 of the depth map 300 through this process.

Figure 11 is a diagram showing a reference graph being formed based on three reference points.

Referring to FIG. 11, the distance calculation unit 161 can select three or more reference objects.

When there are three reference objects, three reference points (P1, P2, P3) are displayed on the distance coordinate plane, and the reference graph (G2) generated based on these may be formed as a curve rather than a straight line.

Since the relative distance and absolute distance to the first reference object are A1 and B1, respectively, the first reference point P1 may be displayed on the distance coordinate plane based on this. Since the relative distance and absolute distance to the second reference object are A2 and B2, respectively, the second reference point P2 may be displayed on the distance coordinate plane based on this. Since the relative distance and absolute distance to the third reference object are A3 and B3, respectively, the third reference point P3 may be displayed on the distance coordinate plane based on this.

As the number of reference objects increases, the reference graph G2 may be generated by reflecting the ratio of the relative distance to the absolute distance for a larger number of reference map objects. When distance transformation is performed using a reference graph (G2) created based on more reference map objects, reliability of the distance transformation may be improved.

Figure 12 is a flowchart showing an image analysis method according to an embodiment of the present invention.

Referring to FIG. 12, in order to calculate the absolute distance of the captured image 200 of the imaging unit 110, which is a monocular camera, the imaging unit 110 may first generate the captured image 200 (S510).

The object detector 140 may detect an object in the captured image 200. Specifically, the object detector 140 may detect a reference object for which standard size information is secured among objects included in the captured image 200.

The depth map generator 150 may generate a depth map for the captured image 200. The distance calculation unit 161 of the map analysis unit 160 may select a reference object from among the reference objects detected by the object detection unit 140. Then, the distance calculation unit 161 calculates the absolute distance of the reference map object 310 using the size of the reference map object 310 corresponding to the reference object, and adds the reference area 410 to the reference map object 310. The relative distance of the reference map object 310 can be calculated by generating .

The distance converter 162 may convert the relative distance for the entire area of the depth map into an absolute distance according to the ratio of the relative distance and absolute distance of at least two reference map objects 310.

Figure 13 is a diagram showing the area of interest being switched.

Referring to FIG. 13, the map analysis unit 160 detects a reference object in a captured image for a preset area of interest 610, 620, and 630 among the captureable area 600 of the imaging unit 110, and detects the reference object. Based on the reference object, the pre-generated absolute distance for the depth map of the captured image can be updated.

As described above, the map analysis unit 160 may calculate the absolute distance for the entire area of the depth map based on the reference object detected in the captured image 200. Meanwhile, the reference object selected by the map analysis unit 160 may be an object that does not have a standard shape and size. In this case, the absolute distance generated based on the corresponding reference object may reflect an error.

In order to compensate for the error reflected in the selection of the reference object, the map analysis unit 160 may update the absolute distance (hereinafter referred to as the total absolute distance) for the entire area of the previously generated depth map.

In order to update the total absolute distance, the map analysis unit 160 can detect a reference object in the captured image for the preset area of interest (610, 620, 630) among the imageable area (600) of the imaging unit (110). there is. Here, the photographable area 600 represents the photographing area of the imaging unit 110 determined by the driving range of the pan and tilt equipment provided in the imaging unit 110.

The map analysis unit 160 may extract a reference map object from a captured image corresponding to the region of interest 610, 620, and 630, and calculate an absolute distance to the extracted reference map object. Additionally, the map analysis unit 160 may update the total absolute distance by applying the absolute distance calculated for the reference map object to the entire area of the depth map. At this time, the map analysis unit 160 may update the total absolute distance when the difference between the newly created total absolute distance and the existing total absolute distance exceeds a preset threshold. The update of the total absolute distance can be continuously performed, and by updating the total absolute distance, the reliability of the total absolute distance for the captured image corresponding to the region of interest 610, 620, and 630 can be improved.

The captureable area 600 of the imaging unit 110 may include a plurality of areas of interest 610, 620, and 630. As shown in FIG. 13, the plurality of regions of interest 610, 620, and 630 may be switched according to a preset order. The areas of interest 610, 620, and 630 are areas that must be specifically observed among the captured images 200, and the user can set the coordinates, conversion time, and conversion order of the plurality of areas of interest 610, 620, and 630 in advance. FIG. 13 shows that the regions of interest 610, 620, and 630 are switched in the order of the first region of interest 610, the second region of interest 620, and the third region of interest 630.

The map analysis unit 160 may detect a reference object in a captured image for a switched interest area 610, 620, and 630 according to a preset order among the plurality of interest areas 610, 620, and 630. If the captured images of the converted regions of interest 610, 620, and 630 include reference objects, the map analysis unit 160 may select a reference object from among the reference objects. Meanwhile, if the captured image of the converted area of interest (610, 620, 630) does not include a reference object, the map analysis unit 160 calculates the total absolute distance based on the area of interest (610, 620, 630). Update can be omitted. Referring to FIG. 13 , when the captured image of the first region of interest 610 includes a reference object, the map analysis unit 160 may update the total absolute distance for the captured image. Next, the object detector 140 may attempt to detect a reference object in the captured image of the second region of interest 620. Thus, when a reference object is detected in the captured image of the second region of interest 620, the map analysis unit 160 may update the total absolute distance for the captured image. Meanwhile, when a reference object is not detected in the captured image of the second region of interest 620, the map analysis unit 160 may omit updating the total absolute distance for the captured image. Subsequently, the object detector 140 may attempt to detect a reference object in the captured image of the third region of interest 630, and this process may be performed for all regions of interest 610, 620, and 630.

Although embodiments of the present invention have been described with reference to the above and the attached drawings, those skilled in the art will understand that the present invention can be implemented in other specific forms without changing the technical idea or essential features. You will understand that it exists. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive.

100: video analysis device 110: imaging unit
120: storage unit 130: control unit
140: object detection unit 150: depth map creation unit
160: map analysis unit 170: output unit

Claims (20)

An imaging unit that generates a captured image;
an object detection unit that detects an object in the captured image;
a depth map generator that generates a depth map for the captured image; and
Select a reference object from among the detected objects, and calculate the absolute distance for the entire area of the depth map by referring to the absolute distance calculated for the reference map object corresponding to the reference object among the map objects included in the depth map. An image analysis device including a map analysis unit. According to claim 1,
The depth map is an image analysis device including relative distance information of objects included in the captured image. According to claim 1,
The object detector detects a reference object for which standard size information is secured among objects included in the captured image,
The map analysis unit is an image analysis device that selects the reference object from the reference objects. According to claim 1,
The map analysis unit,
a distance calculation unit that calculates an absolute distance of the reference map object by referring to the size of the reference map object; and
An image analysis device comprising a distance conversion unit that converts the relative distance for the entire area of the depth map into an absolute distance according to the ratio of the relative distance and absolute distance of at least two reference map objects. According to clause 4,
The distance calculation unit is an image analysis device that calculates an absolute distance of the reference map object by referring to the size of a reference area included in a preset part of the reference map object. According to clause 5,
The distance calculation unit extracts some of the relative distance information included in the reference area and calculates the relative distance of the reference map object. According to clause 6,
The distance calculation unit,
Creating a feature area of a preset size in the reference area,
An image analysis device that determines the average or median value of the relative distances included in the feature area as the relative distance of the reference map object. According to claim 1,
The map analysis unit,
Detecting a reference object in a captured image of a preset area of interest among the imageable areas of the imaging unit,
An image analysis device that updates a pre-generated absolute distance with respect to a depth map of the captured image based on the detected reference object. According to clause 8,
The imageable area includes a plurality of areas of interest,
The map analysis unit is an image analysis device that detects a reference object from a captured image of a region of interest converted according to a preset order among the plurality of regions of interest. According to claim 1,
The object detection unit detects an object in the captured image in real time,
When an object is detected in the captured image according to the real-time detection result, the map analysis unit calculates an absolute distance for the entire area of the depth map. Generating a captured image;
Detecting an object in the captured image;
generating a depth map for the captured image;
selecting a reference object from among the detected objects; and
An image analysis method comprising calculating an absolute distance for the entire area of the depth map by referring to an absolute distance calculated for a reference map object corresponding to the reference object among map objects included in the depth map. According to claim 11,
The depth map is an image analysis method including relative distance information of objects included in the captured image. According to claim 11,
The step of detecting the object includes detecting a reference object for which standard size information is secured among objects included in the captured image,
The step of selecting the reference object includes selecting the reference object from among the reference objects. According to claim 11,
The step of calculating the absolute distance for the entire area of the depth map is,
calculating an absolute distance of the reference map object by referring to the size of the reference map object; and
An image analysis method comprising converting a relative distance for an entire area of the depth map into an absolute distance according to a ratio of the relative distance and absolute distance of at least two reference map objects. According to claim 14,
The step of calculating the absolute distance of the reference map object includes calculating the absolute distance of the reference map object with reference to the size of a reference area included in a preset part of the reference map object. According to claim 15,
An image analysis method further comprising calculating a relative distance of the reference map object by extracting some of the relative distance information included in the reference area. According to claim 16,
The step of calculating the relative distance of the reference map object is,
Creating a feature area of a preset size in the reference area; and
An image analysis method comprising determining an average or median value of relative distances included in the feature area as the relative distance of the reference map object. According to claim 11,
Detecting a reference object in a captured image for a preset region of interest among the captureable areas of the imaging unit that generates the captured image; and
An image analysis method further comprising updating a pre-generated absolute distance with respect to a depth map of the captured image based on the detected reference object. According to clause 18,
The captured image includes a plurality of regions of interest,
The step of detecting a reference object in the region of interest includes detecting a reference object in a captured image of a region of interest converted according to a preset order among the plurality of regions of interest. According to clause 18,
Detecting an object in the captured image includes detecting an object in the captured image in real time,
Calculating the absolute distance for the entire area of the depth map includes calculating the absolute distance for the entire area of the depth map when an object is detected in the captured image according to the real-time detection result. .

Images