Computer vision based wide-area snow/water level estimation using disparity map
Computer vision based, wide-area snow/water level estimation methods using disparity maps. In one embodiment our method provides rich depth-information using a stereo camera and image processing. Scene images at normal and snow/rain weather conditions are obtained by a double-lens stereo camera and a disparity map is generated from the scene images at left and right lenses using a self-supervised deep convolutional network. In another embodiment, our method uses a single point snow/water level sensor, a stationary monocular camera to measure snow/water levels covering a wide area.
1. A computer vision based wide-area snow/water level estimation method using disparity maps, the method:
providing a stereo camera at an elevated height h from the ground, the stereo camera configured to generate images of a scene;
operating the stereo camera during normal and snow/rain weather conditions and collecting images of the scene generated during the normal and snow/rain weather conditions;
using a deep neural network, generating a disparity map that reveals 2D pixel-level depth relation information for the scene;
transforming the disparity map depth-relation information to an absolute distance between the stereo camera and the surface of the ground;
continuously operating the stereo camera during snow/rain weather conditions to generate inclement weather images; and
determining the snow/water level at locations in the inclement weather images.
2. The method according to claim 1 wherein the deep neural network is a self-supervised deep learning neural network.
3. The method according to claim 2 wherein the stereo camera includes a left and a right lens and both images generated at the left and right lens are used as input to a model to generate a detailed disparity map.
4. A computer vision based wide-area snow/water level estimation method using disparity maps, the method:
providing a monocular camera at an elevated height h from the ground, the monocular camera configured to generate images of a scene;
providing a point snow/water level sensor located within the scene imaged by the monocular camera;
operating the camera during normal and snow/rain weather conditions and collecting images of the scene generated during the normal and snow/rain weather conditions, and simultaneously operating the point snow/water level sensor during the normal and snow/rain weather conditions
using a deep neural network, generating a disparity map for the scene;
transforming the disparity map depth-relation information to an absolute distance between the stereo camera and the surface of the ground;
continuously operating the monocular camera during snow/rain weather conditions to generate an inclement weather image; and
determining, from the disparity map and a depth measured by the point snow/water level sensor, the snow/water level at every location in the inclement weather image.
5. The method according to claim 1 wherein the deep neural network is a self-supervised deep learning neural network.