Method for detecting a block raster
View Patent ↗A method for detecting a block raster in an image comprising a number of pixels, each of which having assigned at least one pixel value, said pixels being arranged one after the other along horizontal and vertical pixel boundaries. The method includes establishing a first raster with raster limits (x j +dx j ) that run either parallel to the vertical pixel boundaries or to the horizontal pixel boundaries and, the position of said limits being predetermined by way of an offset (OF) and an opposite distance (SP). A raster scale is determined of the raster as a function of the edge scales of at least some of the raster limits (x j ,+dx j ). The method is repeated for a second raster, and the first or second raster is selected in consideration of the raster scales.
1. A method for the detection of a block raster in an image which has a plurality of pixels, to each of which at least one pixel value is assigned and which are adjacent to one another along horizontal and vertical pixel boundaries (x j , x j +1) the method comprising:
a) establishing a first raster having raster boundaries (x j +dx j ), which run either parallel to the vertical pixel boundaries or parallel to the horizontal pixel boundaries, and the position of which is specified by an offset (OF) and a spacing (SP) between them;
b) ascertaining a raster dimension that is a function of edge dimensions of at least some of the raster boundaries (x j +dx j );
determining of an edge dimension on a raster boundary that lies between two pixel boundaries comprising the following steps
b1) ascertaining an edge dimension of a first pixel boundary which is next to the raster boundary using a high-pass filtering in a direction perpendicular to the first pixel boundary;
b2) ascertaining an edge dimension of a second pixel boundary which is located next to the raster boundary using a high-pass filtering in a direction perpendicular to the second pixel boundary;
b3) weighting the edge dimensions of the first and second pixel boundaries as a function of the spacing between the raster boundary and the first pixel boundary and the second pixel boundary;
c) repeating the method steps a) and b) for a second raster;
d) selecting one of the first and second rasters taking into account the ascertained edge dimensions thereof.
2. The method as recited in claim 1 , in which in step b), the edge dimensions of at least some of the raster boundaries are added together to ascertain the raster dimension.
3. The method as recited in claim 1 , in which in step b) the edge dimension to a first pixel boundary which is located on a first side next to the raster boundary is ascertained and in the same manner to a second pixel boundary which is located next to the raster boundary on a second side opposite to the first side.
4. The method as recited in claim 3 , weighting the edge dimensions produces a weighting factor that decreases in a linear manner with the distance between the raster boundary and a respective pixel boundary.
5. The method as recited in claim 1 , in which the high-pass filtering of steps b1) and b2) each comprise:
selecting pixel groups, each of which has at least two pixels, one of which is located on a first side next to a pixel boundary and another of which is located next to a pixel boundary on a second side which is opposite to the first side;
using a high-pass filtering on the pixel of each of the selected pixel groups to obtain a high-pass filter value for each of them; and
adding the absolute amounts of the individual high-pass filter values.
6. The method as recited in claim 5 in which each pixel group comprises two pixels, a first of which is adjacent to the pixel boundary on the first side and a second of which is adjacent to the pixel boundary on the second side.
7. The method as recited in claim 1 , in which the method steps a) to d) are each performed in sequence at least twice.