Image data processing method by reducing image noise, and camera integrating means for implementing said method

A method of processing vector valued image data by reducing multispectral image noise comprising:

• a) obtaining points of an image with respective vector values associated with the points,
• b) associating a learning zone, comprising a plurality of points, with a reference point of the image, and
• c) assigning the reference point new vector values obtained by a weighted statistical estimation, of weighted mean or weighted median type using the vector values of the points included in the learning zone, wherein
• c1) for each current point of the learning zone, evaluating a distance between a window around the reference point and a window around the current point based on a resemblance between
• the vector values of the points in the reference point window, centered on the reference point and having a size in number of points greater than or equal to 3×3, and
• the vector values of the points in the current point window, of the same format and size as the reference point window and centered on the current point, and
• wherein said resemblance depends on the resemblance between the corresponding vector values of the corresponding points in said reference point window and current point window,
• c2) repeating step c1) for the points of the learning zone in the guise of successive current points by using the distances obtained for all the current points to calculate the weights used in the weighted statistical estimation of step c).