Groups the object candidate rectangles

rects = cv.groupRectangles(rects)
[rects,weights,levelWeights] = cv.groupRectangles(...)
[...] = cv.groupRectangles(..., 'OptionName', optionValue, ...)

Input

• rects Input cell array of rectangles, where each rectangle is represented as a 4-element vector {[x,y,w,h], ...}, or a numeric Nx4/Nx1x4/1xNx4 array.

Output

• rects Output array of retained and grouped rectangles. Same format as input rects (either Nx4 numeric array or cell array). Grouped rectangles are the average of all rectangles in that cluster.
• weights optional output of filled/updated weights, with same length as output rects. See Weights option. Corresponding grouped weights are the maximum weights of all rectangles in that cluster.
• levelWeights optional output of filled/updated level weights, with same length as output rects. See LevelWeights option.

Options

• Thresh Minimum possible number of rectangles in a group minus 1. The threshold is used in a group of rectangles to decide whether to retain it or not. If less than or equal to zero, no grouping is performed. default 1 (i.e only groups with two or more rectangles are kept).
• EPS Relative difference between sides of the rectangles to merge them into a group. default 0.2
• Weights optional vector of associated weights of same length as input rects. Not set by default
• LevelWeights optional vector of doubles of same length as input rects. Not set by default

The function is a wrapper for the generic partition function. It clusters all the input rectangles using the rectangle equivalence criteria that combines rectangles with similar sizes and similar locations. The similarity is defined by EPS. When EPS=0, no clustering is done at all. If EPS -> +inf, all the rectangles are put in one cluster. Then, the small clusters containing less than or equal to Thresh rectangles are rejected. In each other cluster, the average rectangle is computed and put into the output rectangle list. The function also filters out small rectangles inside larger ones that have more hits.

The function is useful for non-max suppression of object detections.