Image processing device for drone, drone image processing method, and drone image processing processor
An image processing device for drone includes a memory that receives and stores image data and a processor that performs pre-processing on the image data by separating the image data into thermal image data and real image data and performs image processing by generating pixel comparison image data as a result of comparing pixel data of the thermal image data and pixel data of the real image data with data of a reference image, correcting data corresponding to an edge of a pixel comparison image in the pixel comparison image data, extracting a change point from the pixel comparison image data that has undergone the correction, and determining whether a change in the change point is in a normal category.
1. An image processing device for a drone, the image processing device comprising:
a memory configured to receive and store image data; and
a processor configured to:
perform pre-processing on the image data by separating the image data into thermal image data and real image data; and
perform image processing by generating pixel comparison image data as a result of comparing pixel data of the thermal image data and pixel data of the real image data with data of a reference image, correcting data corresponding to an edge of a pixel comparison image in the pixel comparison image data, extracting a change point from the pixel comparison image data that has undergone the correction, and determining whether a change in the change point is in a normal category.
2. The image processing device of claim 1 , wherein, during the pre-processing, the processor is further configured to put an alignment reference mark on a structure in the reference image and align a thermal image that corresponds to the thermal image data with a real image that corresponds to the real image data, based on the alignment reference mark.
3. The image processing device of claim 1 , wherein, during the pre-processing, the processor is further configured to set a region of interest and to set a non-region of interest.
4. The image processing device of claim 3 , wherein, during the pre-processing, the processor is further configured to mask the non-region of interest.
5. The image processing device of claim 1 , wherein, during the image processing, the processor is further configured to remove noise by performing Gaussian blurring on the thermal image data and the real image data.
6. The image processing device of claim 1 , wherein, during the image processing, the processor is further configured to convert the thermal image data and the real image data into a color mode and generate the pixel comparison image data, based on a first pixel value difference between the data of the reference image and converted thermal image data and a second pixel value difference between the data of the reference image and converted real image data.
7. The image processing device of claim 6 , wherein, during the image processing, the processor is further configured to determine the color mode based on user customization.
8. The image processing device of claim 1 , wherein, during the image processing, the processor is further configured to determine candidate change points, identify a thermal image part and a real image part of the pixel comparison image, each of the thermal image part and the real image part including a candidate change point having a highest pixel value among the candidate change points, mask a local part of the pixel comparison image, the local part corresponding to the thermal image part and the real image part, compare the candidate change point with the data of the reference image, and extract the change point based on a comparison result.
9. The image processing device of claim 1 , wherein, during the image processing, the processor is further configured to determine, as a candidate change point, a point at which a difference between each of the thermal image data and the real image data and the data of the reference image is greater than an average pixel value of the reference image, compare a thermal data value and a real data value of the candidate change point with a data value of the reference image by using a K-means algorithm, and determine the change point based on a comparison result.
10. The image processing device of claim 1 , further comprising:
a charge station configured to exchange the image data with the drone; and
a data storage that stores the image data,
wherein the processor is mounted on the charge station.
11. An image processing method performed by a drone, the image processing method comprising:
capturing an external image to obtain image data; and
processing the image data,
wherein the processing of the image data includes:
performing pre-processing on the image data by separating the image data into thermal image data and real image data; and
performing image processing by generating pixel comparison image data as a result of comparing pixel data of the thermal image data and pixel data of the real image data with data of a reference image, correcting data corresponding to an edge of a pixel comparison image in the pixel comparison image data, extracting a change point from the pixel comparison image data that has undergone the correction, and determining whether a change in the change point is in a normal category.
12. The image processing method of claim 11 , wherein the performing the pre-processing includes putting an alignment reference mark on a structure in the reference image and aligning a thermal image that corresponds to the thermal image data with a real image that corresponds to the real image data, based on the alignment reference mark.
13. The image processing method of claim 11 , wherein the performing of the pre-processing includes setting a region of interest and setting a non-region of interest.
14. The image processing method of claim 13 , wherein the performing of the image processing includes masking the non-region of interest.
15. The image processing method of claim 11 , wherein the performing the image processing includes removing noise by performing Gaussian blurring on the thermal image data and the real image data.
16. The image processing method of claim 11 , wherein the performing of the image processing includes converting the thermal image data and the real image data into a color mode and generating the pixel comparison image data, based on a first pixel value difference between the data of the reference image and converted thermal image data and a second pixel value difference between the data of the reference image and converted real image data.
17. The image processing method of claim 16 , wherein the performing of the image processing includes determining the color mode based on user customization.
18. The image processing method of claim 11 , wherein the performing of the image processing includes determining candidate change points, identifying a thermal image part and a real image part of the pixel comparison image, each of the thermal image part and the real image part including a candidate change point having a highest pixel value among the candidate change points, masking a local part of the pixel comparison image, the local part corresponding to the thermal image part and the real image part, comparing the candidate change point with the data of the reference image, and extracting the change point based on a comparison result.
19. The image processing method of claim 11 , wherein the performing of the image processing includes determining, as a candidate change point, a point at which a difference between each of the thermal image data and the real image data and the data of the reference image is greater than an average pixel value of the reference image, comparing a thermal data value and a real data value of the candidate change point with a data value of the reference image by using a K-means algorithm, and determining the change point based on a comparison result.
20. A drone image processing processor configured to:
perform pre-processing on image data by separating the image data into thermal image data and real image data; and
perform image processing by generating pixel comparison image data as a result of comparing pixel data of the thermal image data and pixel data of the real image data with data of a reference image, correcting data corresponding to an edge of a pixel comparison image in the pixel comparison image data, extracting a change point from the pixel comparison image data that has undergone the correction, and determining whether a change in the change point is in a normal category.