Apparatus and method for verifying optical fiber work using artificial intelligence
A verification apparatus is provided. The verification apparatus includes: an acquisition unit configured to acquire a photographed image of an optical line for which a splicing operation is completed; an analysis unit configured to extract a defect of a spliced portion of the optical line through analysis of the image; and a determination unit configured to determine the presence or absence of an abnormality in the spliced portion based on the extracted defect.
1 . A verification apparatus comprising:
an acquisition unit configured to acquire a photographed image of an optical line for which a splicing operation is completed;
an analysis unit configured to:
extract a defect of a spliced portion of the optical line through analysis of the image; and
analyze a similarity between each of defect images stored by type in a database and the image acquired by the acquisition unit; and
a determination unit configured to:
analyze a correlation between a plurality of similarities analyzed by type using deep learning; and
determine the presence or absence of an abnormality in the spliced portion based on the correlation.
2 . The verification apparatus of claim 1 , further comprising a splicer including a body part, a guide part, a splicing unit, a cover part, a photographing unit, and a providing unit,
wherein the body part is formed in a three-dimensional shape,
wherein the guide part and the splicing unit are formed on one surface of the body part,
wherein the guide part serves to align a first optical fiber and a second optical fiber so that an end of the first optical fiber and an end of the second optical fiber face each other,
wherein the splicing unit serves to splice the end of the first optical fiber and the end of the second optical fiber aligned to face each other by the guide part,
wherein the cover part is rotatably installed at the body part, and opens or covers the guide part and the splicing unit through the rotation,
wherein the photographing unit is provided on an inner surface of the cover part facing the splicing unit or on one surface of the body part facing the inner surface of the cover part, in a closed state in which the cover part covers the guide part and the splicing unit,
wherein the photographing unit photographs the end of the first optical fiber and the end of the second optical fiber, which become the spliced portion of the optical line, in the closed or an open state,
wherein the providing unit communicates with the acquisition unit through a wired/wireless communication network,
wherein the acquisition unit is provided in a server formed separately from the splicer performing the joining operation and is configured to receive an image of the spliced portion photographed by the photographing unit from the providing unit.
3 . The verification apparatus of claim 1 , wherein
the acquisition unit is configured to acquire an image of a first resolution,
the image of the first resolution is an image photographing a bare fiber exposed to the outside which includes a core and a clad of the optical line,
the image of the first resolution includes a spliced portion of the bare fiber, wherein the verification apparatus further comprises:
a pre-processing unit configured to process the image of the first resolution is provided, wherein
the pre-processing unit is configured to resize the image of the first resolution to a second resolution higher than the first resolution,
the pre-processing unit is configured to extract only a region of the core in which the clad is excluded from the image of the second resolution, and
the analysis unit is configured to analyze the region of the core extracted by the pre-processing unit, and extracts the defect through the analysis.
4 . The verification apparatus of claim 3 , wherein
the pre-processing unit or the analysis unit is configured to generate an enlarged image in which the spliced portion in the region of the core is cropped and enlarged, and
the analysis unit is configured to extract the defect through the analysis of the enlarged image.
5 . The verification apparatus of claim 1 , wherein
the analysis unit is configured to generate a current hash value of the image, and
the analysis unit is configured to extract the defect of the spliced portion by comparing a defect hash value stored in a database with the current hash value.
6 . The verification apparatus of claim 1 , wherein
when different set ranges are set by type of defect,
the determination unit is configured to determine whether a similarity analyzed for each type satisfies a set range for the corresponding type, and
the determination unit is configured to determine the spliced portion as normal only when the similarities analyzed by type satisfy the set ranges for each type, respectively.
7 . The verification apparatus of claim 1 , wherein
the determination unit is configured to determine the presence or absence of an abnormality in the spliced portion using a determination model, and
the determination model is machine-learned using learning data in which a plurality of similarities analyzed by type are taken as a problem and the presence or absence of an abnormality in the spliced portion is an answer.
8 . The verification apparatus of claim 1 , wherein
the determination unit is configured to analyze a correlation between a plurality of similarities analyzed by type,
the determination unit is configured to adjust a limit value of a set range for each type of defect using the correlation,
the determination unit is configured to determine the presence or absence of an abnormality in the spliced portion depending on whether the similarity analyzed for each type satisfies the set range for the corresponding type.
9 . The verification apparatus of claim 1 , wherein
the analysis unit is configured to analyze a first similarity between a first type of defect image and the image acquired by the acquisition unit,
the analysis unit is configured to analyze a second similarity between a second type of defect image and the image acquired by the acquisition unit,
the determination unit is configured to determine the spliced portion as abnormal regardless of the second similarity when the first similarity satisfies a first set value,
the determination unit is configured to set an allowable range of the second similarity to a first range when the first similarity does not satisfy a second set value in a state where the first similarity does not satisfy the first set value,
the determination unit is configured to determine the spliced portion as abnormal when the second similarity does not satisfy the first range in a state where the first similarity does not satisfy both the first set value and the second set value,
the determination unit is configured to set the allowable range of the second similarity to a second range that is smaller than the first range when the first similarity satisfies the second set value in a state where the first similarity does not satisfy the first set value, and
the determination unit is configured to determine the spliced portion as abnormal when the second similarity does not satisfy the second range in a state where the first similarity does not satisfy the first set value and satisfies the second set value.
10 . The verification apparatus of claim 1 , wherein
the analysis unit is configured to analyze a geometric fusion-spliced state of an end of a first optical fiber and an end of a second optical fiber forming the spliced portion,
the determination unit is configured to determine the presence or absence of an abnormality in the spliced portion based on the fusion-spliced state.
11 . The verification apparatus of claim 10 , wherein
the analysis unit is configured to acquire an image of an end portion including the end of the first optical fiber and the end of the second optical fiber,
the analysis unit is configured to recognize the end of the first optical fiber and the end of the second optical fiber through the analysis of the end portion image,
the analysis unit is configured to generate a first bounding box including the end of the first optical fiber,
the analysis unit is configured to generate a second bounding box including the end of the second optical fiber, and
the determination unit is configured to analyze the fusion-spliced state using a difference between center coordinates of the first bounding box and center coordinates of the second bounding box.
12 . A verification apparatus comprising:
a work server formed separately from a splicer for performing a splicing operation for an optical line,
wherein the server includes an acquisition unit, an analysis unit, and a determination unit,
the acquisition unit is configured to obtain splice data including a photographed image of the optical line for which the splicing operation is completed from the splicer,
the analysis unit is configured to analyze the splice data by:
extracting a defect of a spliced portion of the optical line through analysis of the splice data; and
analyzing a similarity between each of defect images stored by type in a database and the image acquired by the acquisition unit,
the determination unit is configured to use the analysis result of the splice data to:
analyze a correlation between a plurality of similarities analyzed by type using deep learning; and
determine the presence or absence of an abnormality in the spliced portion based on the correlation,
the determination unit is configured to transmit a rework message instructing rework to a terminal of a worker who operates the splicer or the splicer when the spliced portion is determined as abnormal, and
the determination unit is configured to transmit a completion message indicating completion of the splicing operation to a management server that manages the optical line when it is determined that the spliced portion is normal.
13 . A verification method which is performed by a verification apparatus, the method comprising:
acquiring a photographed image of an optical line for which a splicing operation is completed;
extracting defects of a spliced portion of the optical line by type through analysis of the image;
analyzing a similarity between each of defect images stored by type in a database and the acquired image;
analyzing a correlation between a plurality of similarities analyzed by type using deep learning; and
determining the presence or absence of an abnormality in the spliced portion based on the correlation.