In-vehicle stereo camera
Provided is an in-vehicle stereo camera which enables the continuation of automated driving if failures occur during imaging. A pair of captured images 301, 302 , which is captured by a pair of imaging units so that both contain an overlapping region, are acquired (S 201 ), it is determined whether an abnormal region is present in at least one of the pair of captured images, and if an abnormal region is present, the degree of impact from the abnormal region on an object recognition process of an object recognition unit is diagnosed (S 202 ), and the processing content of the object recognition process is updated according to the degree of impact (S 203 ).
1. An in-vehicle stereo camera, comprising:
a controller configured to:
acquire a pair of captured images captured by a pair of imaging units so that both contain an overlapping region;
perform an object recognition process that recognizes an object using the pair of captured images;
determine whether or not an abnormal region exists in at least one of the pair of captured images and diagnose, when the abnormal region exists, a degree of impact exerted by the abnormal region on the object recognition process, wherein the degree of impact is based on a size and a transparency of a stain in the at least one of the pair of captured images;
change a processing content of the object recognition process according to the degree of impact; and
in response to diagnosing that the abnormal region exists in either one of the pair of captured images and the degree of impact is equal to or higher than a threshold value, perform the object recognition process by monocular vision using only the other of the pair of captured images.
2. The in-vehicle stereo camera according to claim 1 , wherein the controller is further configured to:
perform object recognition by stereoscopic vision using the pair of captured images in response to diagnosing that the abnormal region does not exist in either of the pair of captured images, or diagnosing that the degree of impact is smaller than the threshold value set in advance.
3. The in-vehicle stereo camera according to claim 1 , wherein the controller is further configured to:
perform object recognition by stereoscopic vision using the pair of captured images in response to diagnosing that the abnormal region does not exist in either of the pair of captured images, or diagnosing that the degree of impact is smaller than a threshold value set in advance, and
perform, in response to diagnosing that the abnormal region exists in the one of the pair of captured images and the degree of impact is equal to or higher than the threshold value:
object recognition by stereoscopic vision using an image of a normal region excluding the abnormal region from the one of the pair of captured images, and the other of the pair of captured images; and
object recognition by monocular vision using an image that is the other of the pair of captured images and is of a region corresponding to the abnormal region.
4. The in-vehicle stereo camera according to claim 2 , wherein the controller is further configured to:
perform processing to calculate a parallax of the pair of captured images, calculate distance information in the captured images based on the parallax, and detect an object based on the distance information as the object recognition by stereoscopic vision; and
perform processing to extract a feature point in the captured images and identify an object based on the feature point as the object recognition by monocular vision.
5. The in-vehicle stereo camera according to claim 2 , wherein by changing a use configuration of hardware, switching is performed to either one of a configuration that executes the object recognition process by stereoscopic vision and a configuration that executes the object recognition process by monocular vision.
6. The in-vehicle stereo camera of claim 1 , wherein in determining whether the abnormal region exists, the controller is configured to calculate a difference between a first stain in a first image of the pair of captured images and a second stain in a second image of the pair of captured images, and determine whether one of the first stain or the second stain is attached to one of the pair of imaging units.