Driving assistance apparatus and driving assistance method
A driving assistance apparatus includes a camera disposed on at least one of a front, sides, or a rear of a vehicle and configured to capture an image of at least one of the front, sides, or rear of the vehicle, and a controller configured to process image data provided from the camera, wherein the controller recognizes whether a surrounding environment of the vehicle is a low-illuminance environment, generates corrected image data using a first artificial neural network model trained to generate the corrected image data in which the image data has been corrected based on the image data when recognizing the low-illuminance environment, and performs traveling control of the vehicle based on the generated corrected image data.
1 . A driving assistance apparatus comprising:
a camera associated with a vehicle, the camera configured to generate an image data of a surrounding environment of the vehicle; and
a controller configured to process the image data provided from the camera,
wherein the controller is further configured to:
recognize whether the surrounding environment of the vehicle is in a low-illuminance environment by calculating a noise amount of the image data provided by the camera and determining that the image data is low-illuminance image data when the calculated noise amount of the image data provided by the camera is greater than or equal to a preset critical noise amount;
when the surrounding environment of the vehicle is in the low-illuminance environment, correct the image data of the surrounding environment of the vehicle provided by the camera using an artificial neural network model trained to generate the corrected image data; and
perform control of the vehicle based on the image data of the surrounding environment of the vehicle corrected using the artificial neural network model.
2 . The driving assistance apparatus of claim 1 , wherein the controller is further configured to correct the image data of the surrounding environment of the vehicle provided by the camera using the artificial neural network model such that a brightness value of the image data of the surrounding environment of the vehicle provided by the camera is increased and a noise amount of the image data of the surrounding environment of the vehicle provided by the camera is decreased.
3 . The driving assistance apparatus of claim 1 , wherein the image data corrected using the artificial neural network model comprises image data having a brightness value greater than a brightness value of the image data provided by the camera and a noise amount smaller than a noise amount of the image data provided by the camera.
4 . The driving assistance apparatus of claim 1 , further comprising an illuminance sensor configured to measure the illuminance around the vehicle,
wherein the controller is further configured to compare the illuminance around the vehicle measured by the illuminance sensor with a preset critical illuminance measurement value to determine whether the surrounding environment of the vehicle is in the low-illuminance environment.
5 . The driving assistance apparatus of claim 1 , wherein the controller is further configured to determine whether the illuminance of the image data provided by the camera is lower than or equal to the preset critical illuminance by:
calculating a brightness value of the image data provided by the camera; and
determining whether the image data provided by the camera is the low-illuminance image data based on the calculated brightness value of the image data provided by the camera.
6 . The driving assistance apparatus of claim 5 , wherein the controller is further configured to determine whether the illuminance of the image data provided by the camera is lower than or equal to the preset critical illuminance by:
calculating brightness values for each pixel of the image data provided by the camera;
determining whether a number of pixels of which the calculated brightness values are smaller than or equal to a preset critical brightness value is greater than or equal to a preset critical pixel number; and
determining that the image data is the low-illuminance image data when the number of pixels of which the calculated brightness values are smaller than or equal to the preset critical brightness value is greater than or equal to the preset critical pixel number.
7 . The driving assistance apparatus of claim 1 , wherein the controller is further configured to:
input the image data provided by the camera to the artificial neural network model; and
generate the corrected image data in which brightness and noise of the image data provided by the camera have been corrected using the artificial neural network model.
8 . The driving assistance apparatus of claim 1 , wherein the artificial neural network model is configured to output the corrected image data by performing at least one of demosaicing, denoising, white balancing, colorization, contrast correction, or demoireing on the image data provided by the camera using the image data provided by the camera as an input for training.
9 . The driving assistance apparatus of claim 1 , wherein the controller is further configured to train the artificial neural network model based on training data including first image data and reference image data corresponding to ground truth of the first image data.
10 . The driving assistance apparatus of claim 9 , wherein the controller is further configured to:
input the first image data to the artificial neural network model;
generate first corrected image data in which the first image data has been corrected using the artificial neural network model; and
train the artificial neural network model so that a difference between the first corrected image data and the reference image data converges to a preset minimum critical value.
11 . The driving assistance apparatus of claim 9 , wherein the controller is further configured to generate the training data using another artificial neural network model trained to output the reference image data using the first image data as an input for training.
12 . The driving assistance apparatus of claim 11 , wherein the another artificial neural network model is based on a generative adversarial network (GAN) model or diffusion model.
13 . A driving assistance method comprising:
recognizing whether a surrounding environment of a vehicle is in a low-illuminance environment by calculating a noise amount of the image data provided by the camera and determining that the image data is low-illuminance image data when the calculated noise amount of the image data provided by the camera is greater than or equal to a preset critical noise amount;
when the surrounding environment of the vehicle is in the low-illuminance environment, correcting image data of the surrounding environment of the vehicle using an artificial neural network model trained to generate the corrected image data; and
performing control of the vehicle based on the image data of the surrounding environment of the vehicle corrected using the artificial neural network model.
14 . The driving assistance method of claim 13 , wherein the recognizing of whether the surrounding environment of the vehicle is in the low-illuminance environment includes:
comparing the illuminance measured by an illuminance sensor of the vehicle with a preset critical illuminance measurement value; and
when the illuminance measured by the illuminance sensor is smaller than or equal to the preset critical illuminance measurement value, determining that the surrounding environment of the vehicle is in the low-illuminance environment.
15 . The driving assistance method of claim 13 , wherein the correcting of the image data of the surrounding environment of the vehicle using the artificial neural network model includes:
inputting the image data of the surrounding environment of the vehicle to the artificial neural network model; and
generating the corrected image data in which brightness and noise of the image data of the surrounding environment of the vehicle have been corrected using the artificial neural network model.
16 . The driving assistance method of claim 13 , further comprising:
inputting first image data among training data including the first image data and reference image data corresponding to ground truth of the first image data to the artificial neural network model;
generating first corrected image data in which the first image data has been corrected using the artificial neural network model; and
training the artificial neural network model so that a difference between the first corrected image data and the reference image data converges to a preset minimum critical value.
17 . The driving assistance method of claim 13 , further comprising generating training data used to train the artificial neural network model using another artificial neural network model trained to output reference image data corresponding to ground truth of first image data using the first image data as an input for training.