| 摘要 |
The invention relates to a method for the automatic recognition of objects and structures in materials, whereby, during a learning phase, a reference material is imaged on the N sensor image points (12) of the colour camera, a signal is generated for each sensor image point (12) and a signal range generated for each of the N sensor image points from the signal values generated during the learning phase. For the recognition of small objects and structures lying below the resolution limit for the colour camera, the material for scanning is imaged on the N sensor image points and each signal generated by the imaging at each individual sensor image point is compared with the signal range for said sensor image point. A binary classification image is generated, whereby an image point of the first class is generated when the signal value for the respective sensor image point (12) lies outside the signal range generated during the learning phase and an image point of the second class is generated when the signal value of the respective sensor image point (12) lies within the signal range generated during the learning phase. Small objects and structures lying below the resolution limit can be reliably recognised by means of said method using a simple 1-chip colour camera. |
