LIDAR data based object recognition apparatus and segment merging method thereof

A segment merging method of a LIDAR data based object recognition apparatus, the segment merging method comprising:

• generating, by a processor, a plurality of segments by grouping points of a point cloud acquired from a LIDAR sensor;
• selecting, by the processor, a target segment among the plurality of segments, wherein the selecting of the target segment includes:
• checking, by the processor, a longitudinal length and a lateral length of a segment;
• checking, by the processor, whether the lateral length of the segment satisfies a first condition;
• checking, by the processor, whether the longitudinal length of the segment satisfies a second condition when the lateral length of the segment satisfies the first condition;
• and
• selecting, by the processor, the segment as a tar get segment when the longitudinal length of the segment satisfies the second condition;
• wherein the first condition is that the lateral length of the segment is less than a first reference value corresponding to a vehicle width;
• and
• wherein the second condition is that the longitudinal length of the segment is greater than a second reference value corresponding to a minimum longitudinal length of a small vehicle and less than a third reference value corresponding to a maximum longitudinal length of the small vehicle;
• checking, by the processor, whether others than the target segment are mergeable segments;
• determining, by the processor, whether to merge the target segment and the mergeable segments based on attribute information of the target segment and the mergeable segments;
• and
• merging, by the processor, the target segment and the mergeable segments to output a merged segment.

A segment meriting method of a LIDAR data based object recognition apparatus, the segment merging method comprising:

• generating, by a processor, a plurality of segments by performing a same segmentation process of grouping points of a point cloud acquired from a LIDAR sensor;
• selecting, by the processor, a target segment among the plurality of segments;
• checking, by the processor, whether others than the target segment are mergeable segments;
• determining, by the processor, whether to merge the target segment and the mergeable segments based on attribute information of the target segment and the mergeable segments;
• and
• merging, by the processor, the target segment and the mergeable segments generated by the same segmentation process to output a merged segment;
• wherein the selecting of the target segment includes:
• checking, by the processor, a longitudinal length and a lateral length of a segment;
• checking, by the processor, whether the lateral length of the segment satisfies a first condition;
• checking, by the processor, whether the longitudinal length of the segment satisfies a second condition when the lateral length of the segment satisfies the first condition;
• and
• selecting, by the processor, the segment as a target segment when the longitudinal length of the segment satisfies the second condition;
• wherein the checking of whether the lateral length of the segment satisfies the first condition includes:
• checking, the processor, whether the longitudinal length of the segment satisfies a third condition when the lateral length of the segment does not satisfy the first condition;
• and
• selecting, by the processor, the segment as a target segment when the longitudinal length of the segment satisfies the third condition;
• wherein the third condition is that the longitudinal length of the segment is greater that a fourth reference value corresponds to a minimum longitudinal length of a vehicle larger than a small vehicle and less than a fifth reference value corresponds to a maximum longitudinal length of the vehicle larger than the small vehicle.

A vehicle, comprising:

• a LIDAR sensor;
• and
• an object recognition apparatus configured to recognize an object by merging over-segmented segments based on points acquired from the LIDAR sensor,
• wherein the object recognition apparatus is configured to:
• generate a plurality of segments by grouping points of a point cloud acquired from a LIDAR sensor;
• select a target segment among the plurality of segments by:
• checking a longitudinal length and a lateral length of a segment;
• checking whether the lateral length of the segment satisfies a first condition;
• checking whether the longitudinal length of the segment satisfies a second condition when the lateral length of the segment satisfies the first condition;
• and
• selecting the segment as a target segment when the longitudinal length of the segment satisfies the second condition;
• wherein the first condition is that the lateral length of the segment is less than a first reference value corresponding to a vehicle width;
• and
• wherein the second condition is that the longitudinal length of the segment is greater than a second reference value corresponding to a minimum longitudinal length of a small vehicle and less than a third reference value corresponding to a maximum longitudinal length of the small vehicle;
• check whether segments other than the target segment among the plurality of segments are mergeable segments;
• and
• determine whether to merge the target segment and the mergeable segments based on attribute information of the target segment and the mergeable segments, and to merge the target segment and the mergeable segments to output a merged segment.