Driving assistance information generating method and device, and driving assistance system

A driving assistance information generating device, comprising:

• an image obtaining module for obtaining a depth image acquired by a depth camera and a scene image acquired by an imaging camera, the depth camera and the imaging camera being mounted on a vehicle in a manner of being registered and matched with each other, and a coverage of the depth camera and a coverage of the imaging camera being at least partially overlapped;
• an object detecting module for detecting positions of respective objects appearing in the scene image by using the depth image and the scene image;
• a driving assistance information generating module for generating driving assistance information of the vehicle based on the positions of the respective objects;
• and
• a lane extracting module for generating lane position information by using the scene image,
• wherein the object detecting module detects the positions of the respective objects appearing in the scene image by using the depth image and the scene image in combination with the lane position information,
• wherein the lane extracting module comprises:
• a line extracting module for extracting a line from the scene image, the line including at least one of a straight line and a curve;
• and
• a lane generating module for generating the lane position information based on the extracted line.

A driving assistance system, comprising:

• a depth camera for acquiring a depth image within its coverage;
• an imaging camera for acquiring a scene image within its coverage, the depth camera and the imaging camera being mounted on a vehicle in a manner of being registered and matched with each other, and the coverage of the depth camera and the coverage of the imaging camera being at least partially overlapped;
• at least one processor;
• at least one memory for storing computer program codes which are configured to perform the following steps when being executed by the processor:
• detecting positions of respective objects appearing in the scene image by using the depth image and the scene image;
• and
• generating driving assistance information of the vehicle based on the positions of the respective objects,
• wherein said detecting positions of respective objects appearing in the scene image by using the depth image and the scene image comprises:
• generating lane position information by using the scene image;
• and
• detecting the positions of the respective objects appearing in the scene image by using the depth image and the scene image in combination with the lane position information,
• wherein said generating lane position information by using the scene image comprises:
• predicting a position of an edge in a current frame of scene image by using a convolutional neural network model and based on each frame of scene image, so as to generate an edge map corresponding to the current frame of scene image;
• extracting a line from the edge map by using a Hough voting algorithm, wherein the line includes at least one of a straight line and a curve;
• and
• obtaining the lane position information based on the extracted line by using a Recurrent neural network model to integrate information on a time axis.