Methods for segmenting objects in an image

Abstract

To provide a computer-implemented method for segmenting an object (OBJ) in at least one image (IM) acquired by a camera (CAM).SOLUTION: A method includes: a) a step of computing an edge probabilities image (EPI) based on an image (IM), the edge probabilities image (EPI) comprising, for each pixel of the image, the probability that the pixel is an edge; b) a step of computing a segmentation probabilities image (SPI) based on the image (IM), the segmentation probabilities image (SPI) comprising, for each pixel of the image (IM), the probability that the pixel belongs to the object (IM); and c) a step of computing a binary mask (MAS) of the object (OBJ) based on the edge probabilities image (EPI) and based on the segmentation probabilities image (SPI).SELECTED DRAWING: Figure 3

Images