Three-dimensional reconstruction method using a plenoptic camera

The method of three-dimensional reconstruction of a surface of interest (200), using a plenoptic camera comprising an entry optical system (110), a matrix optical sensor (130) comprising a plurality of pixels, and a matrix (120;

• 220;
• 330;
• 420) of optical elements (121;
• 221;
• 321;
• 421) arranged between the entry optical system and the matrix optical sensor, characterized in that it comprises the following steps:
• determination of the three-dimensional coordinates of a series of points of the object field of the plenoptic camera, so-called sampling points (Pj);
• determination of calibration data, associating with each sampling point (Pj) a set of at least two pixels (pk) of the matrix optical sensor;
• definition of a three-dimensional grid, so-called reconstruction grid, extending in the object field of the plenoptic camera and whereof each point is associated with one or more sampling points (Pj);
• acquisition (401) of at least one image of the surface of interest by means of the plenoptic camera (100);
• from the calibration data and the image of the surface of interest, calculation (402), for each point of the reconstruction grid, of the value of a dissimilarity index (Cn) which is a function of one or more dispersions, each dispersion being representative of distancing between the values of intensity taken on the image of the surface of interest by the pixels associated with the sampling point (Pj) itself associated with said point of the reconstruction grid, or by the pixels associated with one of the sampling points (Pj) themselves associated with said point of the reconstruction grid;
• determination (403) of a three-dimensional distribution of the points of the reconstruction grid, each allocated with their value of the dissimilarity index (Cn);
• three-dimensional reconstruction (404) of the surface of interest.