Obstacle recognition method and apparatus, device, medium and weeding robot

A method for recognizing an obstacle in a lawn so that a weeding robot working in the lawn can selectively avoid the obstacle, the method comprising the steps of:

• determining a candidate obstacle region in a candidate weeding region image according to color information of the candidate weeding region image;
• obtaining contour information of the candidate obstacle region and chrominance information and value information of the candidate weeding region image;
• and
• determining, according to the contour information and the chrominance information, or the contour information, the chrominance information, and the value information, whether there is an obstacle in the candidate weeding region image so that the weeding robot can be driven accordingly,
• wherein the step of obtaining chrominance information of the candidate weeding region image comprises:
• obtaining a minimum chrominance segmentation value and a maximum chrominance segmentation value of the candidate weeding region image according to the color information of the candidate weeding region image, and
• wherein the contour information comprises at least one of range information, roughness information, a proportion of lawn pixel points, and a proportion of chrominance effective pixels;
• the value information comprises at least one of a number of exposed pixels and a number of non-exposed white pixels;
• and
• the determining, according to the contour information and the chrominance information, or the contour information, the chrominance information, and the value information, whether there is an obstacle in the candidate weeding region image comprises:
• if the minimum chrominance segmentation value is less than a first preset chrominance segmentation threshold and the maximum chrominance segmentation value is greater than a second preset chrominance segmentation threshold, determining, according to the contour information and first preset information determination conditions, or the contour information, the value information, and first preset information determination conditions, whether there is an obstacle in the candidate weeding region image;
• or
• if the minimum chrominance segmentation value is greater than or equal to the first preset chrominance segmentation threshold or the maximum chrominance segmentation value is less than or equal to the second preset chrominance segmentation threshold, determining, according to the contour information and second preset information determination conditions, or the value information and second preset information determination conditions, whether there is an obstacle in the candidate weeding region image, and
• wherein the step of determining, according to the contour information and first preset information determination conditions, or the contour information, the value information, and first preset information determination conditions, whether there is an obstacle in the candidate weeding region image comprises:
• determining that there is an obstacle in the candidate weeding region image if the range information is less than a first preset range threshold, the proportion of lawn pixel points is less than a first preset lawn pixel point proportion threshold, and the roughness information is less than a first preset roughness threshold;
• determining that there is an obstacle in the candidate weeding region image if the proportion of chrominance effective pixels is less than a first preset chrominance effective pixel proportion threshold, and the roughness information is less than a second preset roughness threshold;
• determining that there is an obstacle in the candidate weeding region image if the number of exposed pixels is greater than a first preset exposed pixel number threshold, the number of non-exposed white pixels is greater than a first preset non-exposed white pixel threshold, the proportion of chrominance effective pixels is less than a second preset chrominance effective pixel proportion threshold, and the roughness information is less than a third preset roughness threshold;
• or
• determining that there is an obstacle in the candidate weeding region image if the range information is less than a second preset range threshold, the proportion of lawn pixel points is less than a second preset lawn pixel point proportion threshold, the roughness information is less than a fourth preset roughness threshold, the number of exposed pixels is greater than a second preset exposed pixel number threshold, and the number of non-exposed white pixels is greater than a second preset non-exposed white pixel threshold.