VISUAL SYSTEM FOR THREE DIMENSIONALLY RECOGNIZING POSITIONAL RELATIONSHIP OF DISPERSEDLY LOADED BODIES

摘要

PURPOSE:To easily perform three-dimensional recognition by computing three- dimensional coordinate values of respective points from a two-dimensional image obtained by a visual sensor which employs a light cutting method, and carrying out variable density image formation, edge detection, and split images of respective bodies based on the coordinate values. CONSTITUTION:The image of a part which contains a light cutting line when dispersedly loaded objects (not shown in figure) are irradiated with slit light is picked up by a television camera (not shown in figure) and a coordinate value arithmetic means 51 finds three-dimensional coordinate values of respective points from the obtained two-dimensional image value. A variable density information generating means 53 obtains the variable density image of an (xy) plane based on Z coordinate values as variable density information. Then, a binary information generating means 55 separates the background and a contour information generating means 57 obtain the contours of the respective bodies. Then, a separating means 59 obtains split images of the respective bodies, and a label information means 61, an area arithmetic means 63, a body extracting means 65, and a center gravity arithmetic means 67 recognize the three-dimensional positions of the respective bodies.