Algorithm 3 Separation algorithm of Image Processing Module

Input: An image of touching-pigs Output: An image of individually separated pigs // Hole generation Check natural holes from the touching-pigs If any hole is not detected   Generate artificial holes with erosion or skeleton operator // Guideline generation Select two points with the longest distance in each hole Semi-guideline = Connect (two points of each hole) Guideline = Connect (all of the semi-guidelines) // CP generation ConvexHull (shape of the touching-pigs) Make time-series data L and G Find candidates of CPs with time-series data L Find at most two CPs from the candidates with time-series data G // Segmentation using guideline and CPs If the number of final CPs is two   Connect (guideline, concave points) Else if the number of final CPs is one   Connect (guideline, CP, selected point in the searching region) Else if there is no CP   Connect (guideline, the closest point, selected point in the searching region) Return;