Feature detection device, feature detection method, and computer program for detecting feature

1. A feature detection device comprising:

a processor configured to:

determine whether a vehicle is parked, based on at least one of the speed, steering angle, and shifter position of the vehicle,

detect one or more predetermined features from a first image representing a first area around the vehicle in a section of road in which the vehicle is traveling, the first image being generated by a first camera provided on the vehicle,

detect the one or more predetermined features from a second image representing a second area closer to the vehicle than the first area in a section of road in which the vehicle is parked, the second image being generated by a second camera provided on the vehicle;

execute viewpoint transformation on the second image to generate an aerial image, and detect the one or more predetermined features from the generated aerial image;

estimate a positional relationship between the vehicle and a first feature among the one or more predetermined features, based on the position of the first feature in each of two or more of the first images acquired at different timings during travel of the vehicle and the distance traveled by the vehicle between the timings of acquisition of the two or more of the first images,

estimate a real-space position of the first feature, based on the estimated positional relationship,

identify the position of the first feature in the aerial image, based on the positional relationship between the vehicle and the first feature, and

estimate a real-space position of a second feature detected from the aerial image among the one or more predetermined features, based on a positional relationship between the positions of the first feature and the second feature in the aerial image as well as the real-space position of the first feature.

2. The feature detection device according to claim 1 , wherein the processor is further configured to generate data representing the one or more detected predetermined features.

3. A method for detecting a feature, comprising:

determining whether a vehicle is parked, based on at least one of the speed, steering angle, and shifter position of the vehicle;

detecting one or more predetermined features from a first image representing a first area around the vehicle in a section of road in which the vehicle is traveling, the first image being generated by a first image capturing unit provided on the vehicle;

detecting the one or more predetermined features from a second image representing a second area closer to the vehicle than the first area in a section of road in which the vehicle is parked, the second image being generated by a second image capturing unit provided on the vehicle;

executing viewpoint transformation on the second image to generate an aerial image, and detects the one or more predetermined features from the generated aerial image;

estimating a positional relationship between the vehicle and a first feature among the one or more predetermined features, based on the position of the first feature in each of two or more of the first images acquired at different timings during travel of the vehicle and the distance traveled by the vehicle between the timings of acquisition of the two or more of the first images,

estimating a real-space position of the first feature, based on the estimated positional relationship,

identifying the position of the first feature in the aerial image, based on the positional relationship between the vehicle and the first feature, and

estimating a real-space position of a second feature detected from the aerial image among the one or more predetermined features, based on a positional relationship between the positions of the first feature and the second feature in the aerial image as well as the real-space position of the first feature.

4. A non-transitory recording medium that stores a computer program for detecting a feature, the computer program causing a computer to execute a process comprising:

determining whether a vehicle is parked, based on at least one of the speed, steering angle, and shifter position of the vehicle;

detecting one or more predetermined features from a first image representing a first area around the vehicle in a section of road in which the vehicle is traveling, the first image being generated by a first image capturing unit provided on the vehicle;

detecting the one or more predetermined features from a second image representing a second area closer to the vehicle than the first area in a section of road in which the vehicle is parked, the second image being generated by a second image capturing unit provided on the vehicle;

executing viewpoint transformation on the second image to generate an aerial image, and detects the one or more predetermined features from the generated aerial image;

estimating a positional relationship between the vehicle and a first feature among the one or more predetermined features, based on the position of the first feature in each of two or more of the first images acquired at different timings during travel of the vehicle and the distance traveled by the vehicle between the timings of acquisition of the two or more of the first images,

estimating a real-space position of the first feature, based on the estimated positional relationship,

identifying the position of the first feature in the aerial image, based on the positional relationship between the vehicle and the first feature, and

estimating a real-space position of a second feature detected from the aerial image among the one or more predetermined features, based on a positional relationship between the positions of the first feature and the second feature in the aerial image as well as the real-space position of the first feature.