| 摘要 |
1. Method for identifying the closed contours of objects placed on a surface (2) and contained in a visual field, consisting : - in forming a sampled binary image of the visual field by a line and column distribution of adjacent pixels of polygonal shape, located by line (ILIG) and column (ICOL) indices and bearing a first binary value if they belong to the objects or a second binary value different from the first one in the opposite case, so that the contours of objects are formed of a set of elementary segments called transitions occurring in at least two different directions, with respect to each other, and each separating two adjacent pixels with respective different binary values, the pixels bearing the first binary value being said to be connective if they have by degrees a common point or side ; - in analysing the image sequentially point by point and line by line from a window formed by the pixel being analysed and by adjacent pixels situated on the line and the column immediately preceding the line and column of the pixel being analysed, characterized in that the analysis of the image consists, as each pixel (b0 ) is taken into account : a) in determining the transitions present in the window ; b) in processing solely the consecutive transitions with respective different directions present in the window, while taking into account the connectivity existing between the pixels bearing the first binary value so as to associate with each other two so-called endmost transitions of the same contour present in a frontier of the image which is formed by the set of pixels already analysed of the image and having an adjacent pixel not yet analysed, this processing consisting : - in storing the position of the two endmost transitions of the same contour, which are associated with each other in the frontier of the image ; - in determining an object contour end when the two endmost transitions associated with each other join up with each other. |
