Find a part centroid from image moments

What you need

Use a saved binary mask and OpenCV. Start with one isolated object and a known region of interest.

The calculation

c_x = M10 / M00
c_y = M01 / M00

M00 is area or total mask weight; M10 and M01 are first moments. Coordinates are pixels and require M00 > 0.

Worked example

Try it step by step

1. Reject an empty mask and select the intended component using area and location criteria before calculating moments.
2. Compute the centroid with an explicit zero-area check and draw it over the original image for inspection.
3. Confirm the point lies in a usable grasp region with enough clearance for the fingers or suction cup.
4. Transform the accepted point through a validated camera-to-workplane mapping and reject targets outside the allowed process region.

Offline starter code

This snippet processes local data only; it sends no robot commands.

import cv2
mask = cv2.imread("mask.png", cv2.IMREAD_GRAYSCALE)
if mask is None:
    raise FileNotFoundError("mask.png")
M = cv2.moments(mask, binaryImage=True)
if M["m00"] <= 0:
    raise ValueError("No foreground: do not create a robot target")
print(M["m10"] / M["m00"], M["m01"] / M["m00"])

How to check the result

Compare centroid repeatability over repeated captures and inspect difficult parts where the center is not a physical surface.

Common mistake to avoid

Bounding-box center and area centroid are different. Neither automatically finds a stable grasp point for a hollow or irregular part.

Reference reading

Primary references for the underlying models, APIs or application context. The worked numbers and plots above are educational calculations, not results reported by these sources.

• OpenCV — contour features

Read our methods, limitations and safety notes.