Suzhou Cleva Precision Machinery & Technology Co., Ltd.

An image analysis method includes the steps of obtaining a saturation channel image and a value channel image according to an original image; obtaining a relative deviation value image based on the saturation channel image and the value channel image, wherein the relative deviation value image represents relative deviations between saturation and value; extracting several contours from the relative deviation value image; determining target parameter values corresponding to each contour; and Determining according to analysis results of the target parameter values corresponding to each contour whether there is a non-working region in the original image. A related image analysis apparatus, computer device, and computer readable storage medium are also disclosed.

An obstacle recognition method includes 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. A related obstacle recognition apparatus, electronic device, computer-readable storage medium, and weeding robot are also disclosed.

A self-working system, a self-walking device, a control method therefor and a computer-readable storage medium are disclosed, wherein the control method may include the following steps: acquiring a captured image; processing the captured image to acquire a processed image; segmenting the processed image into at least one sub-region; respectively acquiring a representative pixel point PL n of each sub-region; calculating, in the processed image, the number of sub-regions of which the representative pixel point PL n is located below a comparison pixel point P C and marking same as the number N b of special sub-regions, P C being a preset comparison pixel point; if N b ≤1, judging that there is a boundary ( 2 ) or obstacle in the distance of the captured image; if N b >1, judging that there is a boundary ( 2 ) or obstacle in the vicinity of the captured image. The subsequent work of the self-walking device is directly judged by means of machine vision, which is more convenient and also makes the control more sensitive and effective. Furthermore, an erroneous judgement caused by recognition of a boundary ( 2 ) or obstacle on a slope can be effectively avoided, so that the control is more accurate.

An obstacle recognition method includes 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 value information of the candidate weeding region image; and determining, according to the contour information and the value information, whether there is an obstacle in the candidate weeding region image. A related obstacle recognition apparatus, an electronic device, a computer-readable storage medium, and a weeding robot are also disclosed.

Disclosed are a roughness compensation method and system, an image processing device, and a readable storage medium. The roughness compensation method includes the following steps: acquiring a roughness value of a target region in an image and position information of pixels in the target region; averaging the position information of the pixels in the target region to obtain average position information; determining whether the average position information is located in a preset compensation region; and compensating and updating the roughness value when the average position information is located in the preset compensation region; or skipping compensating the roughness value when the average position information is not located in the preset compensation region. In the present disclosure, a determination result of whether the average position information is located in a preset compensation region is used as a basis of compensation, and a compensated roughness value is used for an image processing program such as shadow recognition, which can effectively reduce erroneous determination of shadow recognition results.

An obstacle recognition method includes the steps of obtaining hue information and value information of a candidate weeding region image; generating a target hue histogram of the candidate weeding region image according to the hue information, and obtaining peak information of the target hue histogram; and determining whether there are obstacles in the candidate weeding region image according to the peak information and the value information. Related apparatus, electronic devices, computer readable storage media, and weeding robots are disclosed.

The present disclosure relates to an image segmentation method based on a chrominance component. The method includes the following steps: obtaining a chrominance component of an image; generating a chrominance component histogram according to the chrominance component; determining peaks and troughs in the chrominance component histogram according to a preset chrominance interval and preset peak and trough setting conditions; obtaining a segmentation threshold according to the peaks and the troughs; and segmenting the image into a plurality of regions with different chrominances according to the segmentation threshold. In the present disclosure, the segmentation threshold is obtained according to the peaks and the troughs, the segmentation threshold is dynamically adjusted according to different images, and a fixed segmentation threshold is not used, thereby effectively reducing false segmentation.

A robot pathfinding method includes obtaining a position of a robot and a position of a charging base in a grid map; obtaining a shortest path from the position of the robot to the position of the charging base; determining candidate transfer stations from the position of the robot to the position of the charging base if the shortest path passes through obstacles; determining a first target transfer station corresponding to the position of the robot and a second target transfer station corresponding to the position of the charging base from the candidate transfer stations; obtaining all candidate transfer paths from the first target transfer station to the second target transfer station; taking the candidate path that is shortest and does not pass through obstacles as a target transfer path; and determining a target path from the position of the robot to the position of the charging base according to a path from the position of the robot to the first target transfer station, the target transfer path, and a path from the second target transfer station to the position of the charging base. Related apparatus, devices, and non-transitory computer-readable storage mediums are disclosed.

A grid map-based robot pathfinding method includes obtaining a first grid map and a second grid map, wherein the second grid map is generated by merging grids in the first grid map, and the resolution of the second grid map is lower than that of the first grid map; planning a travel path of a robot from a current position to a preset target point according to the second grid map; and determining an obstacle position if the travel path is impassable, determining a first path of the robot from the current position to the obstacle position based on the second grid map, and determining a second path of the robot from the preset target point to the obstacle position based on the second grid map; planning a transition path from an endpoint of the first path to an endpoint of the second path according to the first grid map; and obtaining a target path of the robot according to the first path, the second path, and the transition path. Related apparatus, robots, and readable non-transitory storage medium are disclosed.