Image data processing method and device therefor

An image data processing method for treating first image data consisting of a plurality of pixels having values corresponding to their gray level, several of which contain noise components, and transforming them into second image data, in which the noise components are reduced or eliminated, comprising:

• a first step for selecting pixels x.sub.i,j, which are processing objects, whose gray level value is to be updated, from said first image data;
• a second step for obtaining candidate values {x'.sub.i,j } for the gray level, for which, by applying a plurality of values previously prepared, indicating the gray level of the pixels, one after another to a function E ({x.sub.i,j }) consisting of a first variable term U.sup.d ({x.sub.i,j }), whose differential with respect to the gray level {x.sub.i,j } of the object pixel is proportional to the Laplacian with respect to the gray level at the position of the object pixel in the image and a second term U.sup.n ({x.sub.i,j }), which is proportional to the square of the gray level of the noise components contained in the object pixel, and for selecting a candidate value of the gray level, which makes a probability, whose value is determined, depending on said function, greatest;
• and
• a third step for judging by a stochastic method whether the value of the gray level at that time for said object pixel should be replaced by said candidate value or not and for effecting the replacement process depending on the judgment result;
• said first to third steps being performed for all the pixels of said first image data to obtain said second image data.

An image data processing system for treating first image data consisting of a plurality of pixels having values corresponding to their gray level, several of which contain noise components, and transforming them into second image data, in which the noise components are reduced or eliminated, comprising:

• a plurality of multistage shift register means, each of which holds a plurality of pixel data and shifts them;
• connecting means for forming a single ring shift register mechanism by connecting said plurality of shift register means in series in the form of a ring;
• a plurality of operating means for performing operations of same type connected with several particular stages in said ring shift register mechanism;
• and
• input means for distributing inputted image data to said plurality of shift register means so that the relative position of the pixels in the inputted image is maintained;
• whereby said operating means perform operations, which make a probability calculated on the basis of the image energy, which is a function of the values of the image data, greatest by the Monte Carlo method.