Estimation of 3D point candidates from a location in a single image

A method for estimation of 3D point candidates from a single image, comprising:

• capturing color information and depth information in the single image;
• calculating a plurality of planes for selected points in the single image, wherein the plurality of planes are discovered via plane fitting, wherein an inlier threshold of each respective plane of the plurality of planes is increased with the square of an average depth of each respective plane;
• enabling ray casting of the single image at the selected points to obtain three dimensional point candidates at each intersection between a plane of the plurality of planes and a virtual ray;
• and
• merging the three dimensional point candidates for each respective selected point to obtain a location estimate for each selected point.

A system, comprising:

• a display;
• an image capture mechanism to generate a single image;
• a memory that is to store instructions and that is communicatively coupled to the image capture mechanism and the display;
• and
• a processor communicatively coupled to the image capture mechanism, the display, and the memory, wherein when the processor is to execute the instructions, the processor is to:
• calculate a plurality of planes for selected points in the single image, wherein the plurality of planes are discovered via plane fitting, wherein an inlier threshold of each respective plane of the plurality of planes is increased with the square of an average depth of each respective plane;
• enable ray casting of the single image at the selected points to obtain three dimensional point candidates at each intersection between a plane of the plurality of planes and a virtual ray;
• and
• merge the three dimensional point candidates for each respective selected point to obtain a location estimate for each selected point.

A tangible, non-transitory, computer-readable medium comprising instructions that, when executed by a processor, direct the processor to:

• capture color information and depth information in a single image;
• calculate a plurality of planes for selected points in the single image, wherein the plurality of planes are discovered via plane fitting, wherein an inlier threshold of each respective plane of the plurality of planes is increased with the square of an average depth of each respective plane;
• enable ray casting of the single image at the selected points to obtain three dimensional point candidates at each intersection between a plane of the plurality of planes and a virtual ray;
• and
• merge the three dimensional point candidates for each respective selected point to obtain a location estimate for each selected point.