Convolutional Image Processing Software : CAMERA-I
Machine vision (generally used whenever there is a need for visual control of a process or product) relies on a wide variety of mathematical methods to perform image segmentation as well as object grasping. Yet, some of these methods tend to reach their limits when confronted to complex shapes or uneven lighting whereas other’s performances are shadowed by high computation time. The methods proposed here, relies on Maxwell equations (usually devoted to describe electromagnetism laws). An image convolution is applied, based on these equations, and used to determine the specific features of a picture in order to obtain a precise segmentation, then grasping points. This new accurate method, allows for the detection of complex shapes as well as holes and handles while requiring no iteration and very little computing time.