Autonomous cruising system, navigational sign identifying method, and non-transitory computer-readable medium
The present disclosure provides an autonomous cruising system capable of improving an accuracy of identifying description of a sign. The autonomous cruising system includes processing circuitry. The processing circuitry acquires a first image including a buoy from a camera installed in a ship. The processing circuitry identifies a position of the buoy inside the first image. The processing circuitry acquires a second image corresponding to a partial area of the first image including the position of the buoy, and the second image is higher in resolution than the first image. The processing circuitry identifies description of a sign of the buoy based on the second image.
1 . An autonomous cruising system, comprising:
processing circuitry configured to:
acquire a first image including a buoy from a camera installed in a ship,
identify a position of the buoy inside the first image;
acquire a second image corresponding to a partial area of the first image including the position of the buoy, the second image being higher in resolution than the first image;
identify a color candidate, a shape candidate, and a lighting pattern candidate of the buoy in the second image; and
identify description of a sign of the buoy based on the color candidate, the shape candidate, and the lighting pattern candidate of the buoy in the second image, wherein
a greater weight is assigned to the lighting pattern candidate relative to the color and shape candidates during nighttime conditions.
2 . The autonomous cruising system of claim 1 , wherein the camera has an optical zoom function, and
wherein the processing circuitry causes the camera to magnify and image a range corresponding to the partial area of the first image to acquire the second image.
3 . The autonomous cruising system of claim 2 , wherein the camera has a pan/tilt function, and
wherein the processing circuitry directs an imaging direction of the camera toward the range corresponding to the partial area of the first image based on the position of the buoy.
4 . The autonomous cruising system of claim 1 , wherein the processing circuitry acquires the second image by increasing the resolution of the partial area of the first image.
5 . The autonomous cruising system of claim 1 , wherein the processing circuitry acquires the second image by causing an auxiliary camera to image the range corresponding to the partial area of the first image, the auxiliary camera being higher in resolution than the camera.
6 . The autonomous cruising system of claim 1 , wherein the processing circuitry:
identifies the position of the buoy inside the first image by using a first learned model, and
identifies the description of the sign of the buoy based on the second image by using a second learned model.
7 . The autonomous cruising system of claim 6 , wherein the first learned model and the second learned model are formed by incorporating a first learned parameter and a second learned parameter into a common inference program, respectively, the first learned parameter and the second learned parameter being different from each other.
8 . The autonomous cruising system of claim 1 , wherein the processing circuitry further identifies a classification of the buoy.
9 . The autonomous cruising system of claim 8 , wherein, when the classification of the buoy is a lateral sign, the processing circuitry identifies whether the description of the sign of the buoy is either one of a port sign and a starboard sign.
10 . The autonomous cruising system of claim 8 , wherein, when the classification of the buoy is a directional sign, the processing circuitry identifies whether the description of the sign of the buoy is either one of a North sign, an East sign, a South sign, and a West sign.
11 . The autonomous cruising system of claim 1 , wherein, when the classification of the buoy is a lateral sign, the processing circuitry identifies whether the description of the sign of the buoy is either one of the port sign and the starboard sign, by using a learned model for lateral signs, and when the classification of the buoy is a directional sign, the processing circuitry identifies whether the description of the sign of the buoy is either one of a North sign, an East sign, a South sign, and a West sign, by using a learned model for directional signs.
12 . The autonomous cruising system of claim 1 , wherein the processing circuitry does not acquire the second image, when the classification of the buoy is any one of an isolated obstruction sign, a safe water area sign, and a special sign.
13 . The autonomous cruising system of claim 1 , wherein, when a plurality of buoys are identified from the first image, the processing circuitry performs the acquisition of the second image and the identification of the description of the sign for each of the plurality of buoys, respectively.
14 . The autonomous cruising system of claim 1 , wherein the processing circuitry is further configured to display a symbol indicative of the description of the sign of the buoy in any one of the first image, an electronic nautical chart, and a radar image based on the description of the sign of the buoy, the position of the buoy inside the first image, and an imaging direction of the camera.
15 . The autonomous cruising system of claim 1 , wherein the processing circuitry is further configured to determine consistency of the description of the sign of the buoy with description of the sign indicated by navigational sign data recorded on an electronic nautical chart, based on the description of the sign of the buoy, the position of the buoy inside the first image, an imaging direction of the camera, and a position of the ship.
16 . The autonomous cruising system of claim 15 , wherein the processing circuitry is further configured to display a determination result of the consistency in any one of the first image, the electronic nautical chart, and a radar image.
17 . The autonomous cruising system of claim 1 , wherein the processing circuitry is further configured to calculate one of a route of the ship and a width of the route based on the position of the buoy inside the first image and an imaging direction of the camera, when the description of signs of a plurality of buoys includes at least two of a port sign, a starboard sign, and a safe water area sign.
18 . The autonomous cruising system of claim 1 , wherein the processing circuitry is further configured to:
detect a heading of the ship, and
set a direction in which the ship is to travel, based on the description of the sign of the buoy, an imaging direction of the camera, and the heading of the ship.
19 . A method of identifying a navigational sign, comprising the steps of:
acquiring a first image including a buoy from a camera installed in a ship;
identifying a position of the buoy inside the first image;
acquiring a second image corresponding to a partial area of the first image including the position of the buoy, the second image being higher in resolution than the first image;
identifying a color candidate, a shape candidate, and a lighting pattern candidate of the buoy in the second image; and
identifying description of a sign of the buoy based on the color candidate, the shape candidate, and the lighting pattern candidate of the buoy in the second image, wherein
a greater weight is assigned to the lighting pattern candidate relative to the color and shape candidates during nighttime conditions.
20 . A non-transitory computer-readable medium having stored thereon computer-executable instructions which, when executed by a computer, cause the computer to:
acquire a first image including a buoy from a camera installed in a ship;
identify a position of the buoy inside the first image;
acquire a second image corresponding to a partial area of the first image including the position of the buoy, the second image being higher in resolution than the first image; and
identify a color candidate, a shape candidate, and a lighting pattern candidate of the buoy in the second image; and
identify description of a sign of the buoy based on the color candidate, the shape candidate, and the lighting pattern candidate of the buoy in the second image, wherein
a greater weight is assigned to the lighting pattern candidate relative to the color and shape candidates during nighttime conditions.