Table 4 A comparison of segmentation techniques

Clustering-based

Segments the image into clusters consisting of pixels with similar characteristics

(1) Elimination of noisy spots

(2) Typically obtains homogeneous regions

(1) Sensitive to noise

(2) Hard to find initial parameters

Color-index-based

Makes a distinction between foreground and background values based on a scalar value (e.g., green channel)

(1) Simple to implement

(2) Low computational cost

(3) High efficiency

(1) Omitting spatial information by only considering pixel intensities

(2) Sensitive to noise

Edge-based

Detects edge points based on sudden changes in intensity and generates edge segments by grouping edge points together

(1) High accuracy in edge positioning

(2) High speed

(1) No guarantees about continuity and closure of edges

(2) Less suitable for images with many edges

Region-based

Divides the point cloud into different clusters based on local smoothness and curvature characteristics or on the presence of features at a certain scale

(1) Effective for complex images

(2) High accuracy in images with high contrast between regions

(3) Generally good performance in noisy images

(1) Complicated algorithm

(2) Computationally intensive

Threshold-based

Divides pixels into groups based on their intensity relative to a given value or threshold

(1) Simple to implement

(2) Low computational cost

(3) High efficiency

(1) Depending only on the pixel gray value without considering spatial details

(2) Sensitive to noise