value Other recorded cells and in their this cell. Their number The is equal scale to ratio K M.
values, which are adjacent to the central cell. The first absorbed Availability cells are the of excitation neighboring signal cells e with i t excite equal. If K M <1, States then of the a cell cells at are time absorbed t for one in of the the direction coordinates to b the i t. At different scaling factors for different coordinates also the image changes in the direction of these coordinates (see figure 7). The reproduction is carried out similarly. When all of the cells and their values on one coordinate proliferate in opposite sides from the cell of scaling center, cells reproduction begins according to another coordinate. The other cells and their values shift on the K M of cells from the cell of scaling center (figure 6). The first cell creates additional cells on one of the coordinates with the same conditions. The closest cells to the central cell first begin to reproduce. If K M > 1 cell reproduction begins with shift on both coordinates in turn. Each cell of CA is given by a scaling factor К М. The method is described below: Initially, a cell is selected, which is the center of scaling. There are many problems in the known image scaling methods that are associated with the assignment of the center of scaling and arbitrary scaling factors for X and Y coordinates. Consequently, all depends on the accurate initial morphological image processing by the camera. The relations obtained show that the difference between the obtained values of the ratio to ratio of real images and the computer images, practically negligible. With the help of these experiments average values of the ratios of the number of cells that form a contour for different types of neighborhoods were obtained. Histograms of cells that form the edges of the figures images for given forms of neighborhood are shown in Figure 5. Image processing was carried out by the program "Study". For existing an image binarization with sensitivity in the range from 50 to 60 percent was carried out. For the contrast selection and binarization of the resulting images boundaries with effective sensitivity were defined where unneeded cells were completely removes, which did not belong to the figure. At the same time each figure was rotated on a certain angle. The resulting image was divided into separate images of the equal dimension (300×300). In the next part of the experimental research real images of figures obtained by photographing were used. The experiments were conducted for symmetric structures of neighborhoods that have the same shape but different in the number of cells that form the neighborhood. At the same time shape of the curves are the same for all types of neighborhoods at the same sample of images, which have been investigated. The analysis of the dependency shows the stability of the difference between the numbers of cells that form the edges. Graph of cell contour distribution for the neighborhood of von Neumann, Moore, of the second and third orders is shown in Figure 4. For all the selected types of neighbors their comparative analysis was performed by number of cells that form edges.
#Object2vr image scaling code
CU compares it with a predetermined threshold value and with code from output of the ME, and accordingly controls the state of ME. Cell contains a memory element (ME), comparing unit (CU) and the adder of mean value (AMV) that summarizes values of the neighborhood cells and determines the average value. Cell structure, which implements the CA for edge detection of the halftone image is submitted on figure 3. Also, this approach allows the edges selection of images by gradations of brightness. Threshold variation value allows to select edge as well as to implement noise filtering, which presented in the image. Otherwise, the cell is set to value of background ( fig. If the value exceeds the control cell threshold, this cells belongs to the cells of the edge. Each cell is determined its the mean value from value set of cells of the neighborhood and is compared with a predetermined threshold value. The CA separates cells of object from the background on the base of selected threshold value of brightness. However, for halftone image processing and edge detection of the object it is necessary to use thresholding processing. Through variation of the neighborhood structure we can create different shapes and transformation of the objects contours.
we can see from the illustrated example it is possible to create a shadow of the object, to make some lines thicker, and the others are thinner.