JP7196490B2 - Living body detection method - Google Patents

Description

The present invention relates to a living body detection method for detecting a living body appearing in a moving image.

2. Description of the Related Art Conventionally, flying objects (also called multicopters or drones) have been used to survey the movement range and population of living organisms. In recent years, the use of flying objects has been studied for investigating animals (vermin) that may harm humans and crops, such as deer, wild boars, and bears.

For example, Patent Literature 1 discloses an animal habitat survey method using a flying object equipped with an infrared thermography camera. In Patent Document 1, a flying object equipped with an infrared thermography camera is caused to fly along a preset flight route, and when it reaches a predetermined shooting position or periodically, the infrared thermography camera shoots, and the latitude is added to the shot image. , longitude, and altitude, and stored in the storage means. Then, in the computer device, a specific animal is detected from the photographed images stored in the storage means, and based on the extracted animal population in a certain range, the animal population in the entire target area is estimated and calculated. there is

Japanese Patent Application No. 2017-194760

In Patent Literature 1, when detecting a specific animal, the individual position is extracted by pattern matching based on the brightness image in the thermo image. With such a method, processing is performed to check whether the luminance image in the thermo image matches the pattern shape, so the processing load is large. In addition, since animals appear extremely small in photographs taken from the sky, it is not possible to acquire their shapes clearly enough for pattern matching. Therefore, the technique of Patent Document 1 has room for further improvement.

SUMMARY OF THE INVENTION In view of such problems, it is an object of the present invention to provide a living body detection method capable of easily distinguishing a living body while reducing the load of the living body detection processing.

In order to solve the above problems, a representative configuration of the living body detection method according to the present invention is a living body detection method for detecting a living body appearing in a moving image, wherein the moving image is captured by an aircraft equipped with an infrared camera. A frame of a still image is extracted from the moving image as an original image, the original image is Fourier transformed to analyze the spatial frequency distribution, and frequency processing is performed to remove a predetermined range of spatial frequencies from the spatial frequency distribution. , performing inverse Fourier transform on the spatial frequency distribution after removal processing to generate an extracted image, performing binarization processing on the extracted image, and superimposing the contour of the region corresponding to the living body on the original image. It is characterized by highlighting a living body.

In the above configuration, an extracted image is generated by Fourier transforming, frequency processing, and inverse Fourier transforming an original image extracted from a moving image. As a result, the living body can be detected without performing pattern matching processing, and the load required for the living body detection processing can be reduced. By superimposing the extracted image on the original image and highlighting the living body, the living body can be easily identified.

It is preferable to perform spatiotemporal median filter processing on the image subjected to the above binarization processing using the original images of the frames before and after the binarization processing. With such a configuration, it is possible to preferably remove noise in the extracted image after the inverse Fourier transform.

The image subjected to the binarization process is subjected to a maximum value filter process in which the maximum value among the luminances in the vicinity of the pixel of interest is used as the new luminance of the pixel of interest, and the region with the maximum luminance is the region corresponding to the living body. should be The image subjected to the above binarization processing is subjected to minimum value filter processing using the minimum value among the luminances in the vicinity of the pixel of interest as the new luminance of the pixel of interest, and the contour of the area with the lowest luminance corresponds to the living body. It is preferable to use the area where As a result, the area of the living body in the extracted image can be enlarged. Therefore, it becomes possible to distinguish the living body more easily.

According to the present invention, it is possible to provide a living body detection method capable of easily distinguishing a living body while reducing the load of living body detection processing.

It is a figure explaining the apparatus used in the living body detection method concerning this embodiment. 4 is a flowchart of a living body detection method according to the present embodiment; FIG. 4 is a diagram showing an example of a processed image; FIG. FIG. 4 is a diagram showing an example of a processed image; FIG. FIG. 4 is a diagram showing an example of a processed image; FIG. FIG. 4 is a diagram showing an example of a processed image; FIG.

Preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The dimensions, materials, and other specific numerical values shown in these embodiments are merely examples for facilitating understanding of the invention, and do not limit the invention unless otherwise specified. In the present specification and drawings, elements having substantially the same functions and configurations are denoted by the same reference numerals to omit redundant description, and elements that are not directly related to the present invention are omitted from the drawings. do.

FIG. 1 is a diagram for explaining an apparatus used in the living body detection method according to this embodiment. In the living body detection method of this embodiment, the arithmetic device 100 and the flying object 200 shown in FIG. 1 are used. The flying object 200 is equipped with an infrared camera 210, and the infrared camera 210 captures moving images. The computing device 100 receives the moving image captured by the infrared camera 210, processes the moving image, and detects the living body L appearing in the moving image. The computing device 100 may be a dedicated device, but more specifically, it can be constructed with a general-purpose computer.

FIG. 2 is a flow chart of the living body detection method according to this embodiment. 3 to 6 are diagrams showing examples of processed images. In the living body detection method of this embodiment, the flying object 200 flies and the living body L appearing in the moving image captured by the infrared camera 210 is detected. Since the flying object 200 can fly and photograph places that are not easily accessible to observers, such as mountainous areas and forests, it is possible to conduct surveys in places that cannot be seen.

In the living body detection method of the present embodiment, as shown in FIG. 2, the arithmetic device 100 acquires a moving image by receiving the moving image photographed by the infrared camera 210 of the aircraft 200 (step S300). Next, the arithmetic device 100 extracts a still image frame from the moving image as an original image (step S302). The extracted original image is in a completely unprocessed state as shown in FIG. 3(a).

Subsequently, the arithmetic device 100 Fourier-transforms the original image shown in FIG. 3A (step S304). Thereby, the distribution of spatial frequencies in the original image is analyzed. Then, the arithmetic device 100 performs frequency processing on the original image that has undergone the Fourier transform processing (step S306). As for the contents of the frequency processing, spatial frequencies in a predetermined range are removed from the spatial frequency distribution. As an example, leaves of trees have a low spatial frequency due to a small temperature difference. Since a living body has a large temperature difference with respect to its surroundings, the difference in brightness also increases on the image, resulting in a high spatial frequency. Therefore, by leaving the spatial frequencies of the band representing the living body and its surroundings and removing the other bands, it is possible to emphasize the contour of the living body.

After that, the arithmetic device 100 performs an inverse Fourier transform on the spatial frequency distribution after removal processing (step S306). As a result, an extracted image as shown in FIG. 3(b) is generated from the spatial frequency. As is clear from FIG. 3B, it can be understood that the background other than the living body can be largely removed by performing the removal processing as described above.

In particular, in the living body detection method of this embodiment, as shown in FIG. 2, binarization processing is performed after performing inverse Fourier transform processing (step S310). Spatio-temporal median filter processing is performed using the original image of the frame (step S312). By performing the binarization process, the extracted image in FIG. 3(b) can be converted into a monochrome image consisting of the living body L and the background other than the living body L as shown in FIG. 4(a) (in this embodiment, black and white is reversed). By performing spatio-temporal median filter processing in addition to binarization processing, noise in the extracted image after the inverse Fourier transform can be preferably removed.

After the binarization process and the spatio-temporal median filter process are performed as described above, the processed image is subjected to maximum value filter process or minimum value filter process (step S314). In the maximum value filtering process, the maximum value among the luminances in the neighborhood of the pixel of interest is used as the new luminance of the pixel of interest, and the outline of the region with the maximum luminance is extracted from the binarized image. In the minimum value filtering process, the minimum value of the luminances in the neighborhood of the pixel of interest is used as the new luminance of the pixel of interest, and the outline of the region with the minimum luminance is extracted from the binarized image.

In the present embodiment, since black and white are reversed in FIG. 4A and the living body is represented in white (with high luminance), maximum value filtering is performed. As a result, the area of the living body L in the image shown in FIG. 4(a) is enlarged as shown in FIG. 4(b). Therefore, it becomes possible to distinguish the living body L more easily. When the living body is expressed in black (low brightness), minimum value filtering may be used.

Then, when the maximum brightness area shown in FIG. 4(b), that is, the outline C of the area corresponding to the living body L is extracted, the state shown in FIG. 5(a) is obtained. After extracting the contour C, the contour C of the extracted image is superimposed on the original image (see FIG. 3A) (step S316). As a result, the living body L is highlighted by the outline C, as shown in FIG. 5(b). The contour C can be appropriately colored in a conspicuous color such as red or orange.

In FIG. 6(a), processing is performed to increase the thickness of the contour C shown in FIG. 5(a). Then, in FIG. 6B, the thickened contour C is superimposed on the original image. As a result, the living body L can be visually recognized more easily than when the contour C is thin.

As described above, according to the living body detection method of the present embodiment, an extracted image is generated by Fourier transforming, frequency processing, and inverse Fourier transforming an original image extracted from a moving image. As a result, the living body can be detected without performing pattern matching processing, and the load required for the living body detection processing can be reduced. By superimposing the extracted image on the original image and highlighting the living body, the living body can be easily identified.

Although the preferred embodiments of the present invention have been described above with reference to the accompanying drawings, it goes without saying that the present invention is not limited to such examples. It is obvious that a person skilled in the art can conceive of various modifications or modifications within the scope described in the claims, and these also belong to the technical scope of the present invention. Understood.

INDUSTRIAL APPLICABILITY The present invention can be used as a living body detection method for detecting a living body appearing in a moving image.

DESCRIPTION OF SYMBOLS 100... Arithmetic device, 200... Flight object, 210... Infrared camera, C... Contour, L... Living body

Claims (4)

A living body detection method for detecting a living body appearing in a moving image,
A moving image is taken by an aircraft equipped with an infrared camera,
extracting a frame of a still image from the moving image as an original image;
Analyzing the spatial frequency distribution by Fourier transforming the original image,
performing frequency processing to remove spatial frequencies in a predetermined range from the spatial frequency distribution;
generating an extracted image by performing an inverse Fourier transform on the spatial frequency distribution after removal processing;
performing a binarization process on the extracted image,
subjecting the binarized image to a maximum value filtering process in which the maximum value among the luminances in the vicinity of the pixel of interest is set as the new luminance of the pixel of interest;
extracting the outline of the region of maximum luminance as the region corresponding to the living body;
A living body detection method, wherein the living body is highlighted by superimposing a contour of a region corresponding to the living body on the original image. A living body detection method for detecting a living body appearing in a moving image,
A moving image is taken by an aircraft equipped with an infrared camera,
extracting a frame of a still image from the moving image as an original image;
Analyzing the spatial frequency distribution by Fourier transforming the original image,
performing frequency processing to remove spatial frequencies in a predetermined range from the spatial frequency distribution;
generating an extracted image by performing an inverse Fourier transform on the spatial frequency distribution after removal processing;
performing a binarization process on the extracted image,
performing minimum value filtering on the image subjected to the binarization process, wherein the minimum value among the luminances in the vicinity of the pixel of interest is set as the new luminance of the pixel of interest;
extracting the outline of the area with the lowest brightness as the area corresponding to the living body;
A living body detection method, wherein the living body is highlighted by superimposing a contour of a region corresponding to the living body on the original image. 2. The method according to claim 1 , wherein the binarized image is subjected to spatio-temporal median filtering using original images of preceding and succeeding frames, and then the maximum value filtering is performed . A living body detection method as described. 3. The method according to claim 2, wherein the binarized image is subjected to spatio-temporal median filtering using original images of preceding and succeeding frames, and then the minimum value filtering is performed. liveness detection method.

Images