Leakage oil detection device and leakage oil detection method
The present invention is directed to a leakage oil detection device for detecting leakage oil in an oil filled apparatus including: a light source configured to irradiate the apparatus; an imaging device configured to capture an image of the apparatus; a control device configured to control operations of the light source and the imaging device; a storage device configured to store a captured image; an image processing device configured to process the stored image; and a display device configured to display a processing result. The light source and the imaging device are arranged to capture specular reflection light from a surface of the apparatus as a target object. The image processing device recognizes a brightest portion and an adjacent dark portion on the image as leakage oil adhesion portions when a three-layer structure of luminance having bright portions having different luminances and a dark portion is observed in the image.
1. A leakage oil detection device configured to detect leakage oil in an oil filled apparatus, the leakage oil detection device comprising:
a light source configured to irradiate the oil filled apparatus with light;
an imaging device configured to capture an image of the oil filled apparatus;
a control device configured to control operations of the light source and the imaging device;
a storage device configured to store a captured image;
an image processing device configured to process the stored image; and
a display device configured to display a processing result, wherein
the light source and the imaging device are arranged in a manner of being capable of capturing specular reflection light from a surface of the oil filled apparatus as a target object,
the image processing device
recognizes a brightest portion and a dark portion adjacent to the brightest portion on the captured image as leakage oil adhesion portions when a three-layer structure of luminance having bright portions having different luminances and a dark portion is observed in the image, and
extracts only a leakage oil site by machine learning method with respect to the observed three-layer structure of luminance, and
the display device displays by adding a marker to an original image or the same image obtained without irradiation of the light source.
2. The leakage oil detection device according to claim 1 , wherein
light of the light source includes both ultraviolet light and visible light,
a filter that transmits the ultraviolet light and the visible light is mounted in front of the light source, and
a filter that transmits visible light is mounted in front of the imaging device.
3. A leakage oil detection device configured to detect leakage oil in an oil filled apparatus, the leakage oil detection device comprising:
a light source configured to irradiate the oil filled apparatus with light;
an imaging device configured to capture an image of the oil filled apparatus;
a control device configured to control operations of the light source and the imaging device;
a storage device configured to store a captured image;
an image processing device configured to process the stored image;
a display device configured to display a processing result, wherein
the light source and the imaging device are arranged in a manner of being capable of capturing specular reflection light from a surface of the oil filled apparatus as a target object, and
the image processing device
recognizes a bright portion in luminance as a leakage oil adhesion portion when the bright portion is observed inside a dark portion of the captured image at an end portion of the image, and
extracts only a leakage oil site by machine learning method with respect to the observed three-layer structure of luminance, and
the display device displays by adding a marker to an original image or the same image obtained without irradiation of the light source.
4. A leakage oil detection method for detecting leakage oil of an oil filled apparatus by using an image obtained by irradiating the oil filled apparatus with light and capturing an image of the oil filled apparatus, the leakage oil detection method comprising:
recognizing a brightest portion and a dark portion adjacent to the brightest portion on the captured image as leakage oil adhesion portions when a three-layer structure of luminance having bright portions having different luminances and a dark portion is observed in the image,
extracting only a leakage oil site by machine learning method with respect to the observed three-layer structure of luminance, and
adding a marker to an original image or the same image obtained without irradiation of the light source.
5. A leakage oil detection method according to claim 4 , the leakage oil detection method comprising:
recognizing a bright portion in luminance, being different than the brightest portion, as a leakage oil adhesion portion when the bright portion is observed inside a dark portion of the captured image at an end portion of the image.