Headlight control device, headlight control system, and headlight control method

1 . A headlight control device that controls light distribution of a headlight mounted on a vehicle, the headlight control device comprising:

the headlight being configured to change a light distribution pattern;

processing circuitry:

to calculate visual saliency on a basis of luminance, color or direction of imaging signals indicating ahead of the vehicle by applying deep learning including a neural network, and generate a saliency map indicating a distribution pattern of a gaze region that is a region gazed by a driver of the vehicle on a basis of the calculated visual saliency;

to determine a scene that is a situation in which the vehicle is traveling on a basis of a distribution pattern of gaze regions indicated by the saliency map generated by calculating probability density distribution of gazes in a horizontal direction and a vertical direction on a basis of the visual saliency map and by evaluating features of the calculated probability density distribution; and

to determine the light distribution pattern of the headlight on a basis of the scene determined, and output a light distribution control signal for controlling the headlight to emit light in the determined light distribution pattern,

wherein the headlight is controlled such that the light distribution pattern is changed based on the output light distribution control signal to irradiate an object in the gaze region with light.

2 . The headlight control device according to claim 1 ,

wherein the processing circuitry specifies a compatible region that is a region where an obstacle does not exist and is the gaze region ahead of the vehicle on a basis of an obstacle detection signal related to the obstacle with respect to the vehicle and the saliency map generated, and determines the scene on a basis of a distribution pattern of the specified compatible region.

3 . The headlight control device according to claim 2 ,

wherein

the processing circuitry further specifies a near miss region, which is a region where a pedestrian is likely to jump out, on a basis of a specified position of the compatible region and a position of the obstacle indicated by the obstacle detection signal, and

the processing circuitry determines the light distribution pattern further on a basis of the near miss region specified.

4 . The headlight control device according to claim 2 ,

wherein the processing circuitry determines that the scene is a wet road surface in a case where the specified compatible region exists on a road surface.

5 . The headlight control device according to claim 1 ,

wherein

the processing circuitry determines whether or not the scene is a slope having a gradient on a basis of displacement in a vertical direction of a gaze region indicated by the saliency map generated, and

when the processing circuitry determines that the scene is the slope, the processing circuitry adjusts the light distribution pattern so as to move the light emitted from the headlight in the vertical direction.

6 . The headlight control device according to claim 1 ,

wherein the processing circuitry determines that the scene is a general road, an expressway, or an urban area on a basis of a distribution pattern in a horizontal direction of a gaze region indicated by the saliency map generated.

7 . The headlight control device according to claim 1 ,

wherein the processing circuitry generates the saliency map on a basis of an infrared captured image as the imaging signal.

8 . The headlight control device according to claim 1 ,

wherein the processing circuitry generates the saliency map further on a basis of a position of the vehicle and three-dimensional map information.

9 . The headlight control device according to claim 8 ,

wherein the processing circuitry generates the saliency map further on a basis of a past saliency map generated in association with a position of the vehicle in the past.

10 . The headlight control device according to claim 1 ,

wherein the processing circuitry generates the saliency map further on a basis of vehicle speed information indicating a speed of the vehicle.

11 . A headlight control system comprising:

a headlight control device according to claim 1 ;

an imager to acquire an image indicating ahead of the vehicle by imaging the front of the vehicle, and convert the acquired image into the imaging signal; and

a headlight to emit light to the front of the vehicle in a light distribution pattern indicated by a light distribution control signal output.

12 . A headlight control method performed by a headlight control device that controls light distribution of a headlight, configured to change a light distribution pattern, mounted on a vehicle, the headlight control method comprising:

calculating visual visible saliency on a basis of luminance, color or direction of imaging signals indicating ahead of the vehicle and generating by applying deep learning including a neural network, a saliency map indicating a distribution pattern of a gaze region that is a region easily gazed by a driver of the vehicle on a basis of the calculated visual saliency;

determining a scene that is a situation in which the vehicle is traveling on a basis of a distribution pattern of gaze regions indicated by the saliency map generated by calculating probability density distribution of gazes in a horizontal direction and a vertical direction on a basis of the visual saliency map and by evaluating features of the calculated probability density distribution; and

determining the light distribution pattern of the headlight on a basis of the scene determined, and outputting a light distribution control signal for controlling the headlight to emit light in the determined light distribution pattern,

wherein the headlight is controlled such that the light distribution pattern is changed based on the output light distribution control signal to irradiate an object in the gaze region with light.