| Popis: |
The operation of algorithms for identifying object’s contours using two-dimensional convolution operation on the boundaries of raster images depends entirely on the location of the transformed pixel in the pixel matrix. The problem is related to the need for the convolution algorithm itself to use an additional set of pixels located outside the pixel matrix of the processed image [1, 2, 3]. In this paper, an algorithm for balanced interpolation of pixels at image boundaries in addition to algorithms based on statistical averages [4, 5] is proposed. Various sets of coefficients which are located at a certain pixel layout and participate in the interpolation algorithm is considered. A comparative analysis of the results of the balanced interpolation algorithm with the results using the arithmetic mean methods is carried out. Also, an example of using a new algorithm to solve the problem of the object’s contours identification on the borders of a raster halftone image is presented. The described algorithm is easy to understand, use and integrate with other algorithms and it does not require significant computing power. |
