TW416231B - Zooming method for digital image - Google Patents

Abstract

The present invention is a zooming method for digital image which provides a method of using area mapping to implement uniform and smooth zooming of the image. The method employs the mapping relationship between pixel coordinates from the original image to the new image so as to get the area of pixel coordinates surrounding the new image; then, applying averaging or interpolation in this area to construct each pixel of the new image and achieve the uniform and smooth zooming effect for the digital image processing on the computer and device with LCD. The method will not generate mosaic effect while zoom-in the screen image and will not generate distortion during zoom-out and also provide high speed for processing zooming.

Description

416231

V. Description of the invention (i) The method of the scope of the invention, and particularly the method of attacking the moon beam or having an L CD. The invention relates to a kind of digital image processing which is a kind of processing applied to the screen image, which can make the digital image processing of the electric device have uniform and smooth shrinkage. [Background of the invention] When the image on the computer screen is processed, it needs to be processed. In terms of reduction or enlargement (raster image), for example, the screen image is combined by color to produce a complete number of primes. It is fixed and limited, so it is possible to put it at the same resolution. The generation method is by uniformly deleting or interpolating. If the image is enlarged or reduced, or pixels are deleted without further mosaic or distortion, so 'for grayscale or full-color (physical) traditional scaling methods are generally Interpolation and high-order interpolation methods are used to improve the conversion of images to high frequencies. It is difficult to avoid mosaics, but the operation is slow. [Summary of the Invention] Too large or too + is not conducive to display or large. For dot matrix images The image is composed of many pixels (Pixel light spots, each pixel's color bit image °) Because the image constituting the image is small or rotated, it will affect the image. Distortion. Generally, the digital image is scaled into the scene; the pixels of the image are used to achieve it. However, if the nearest image is simply inserted, the resulting image will have a real phenomenon. Fu 1 1 c ο 1 〇r) The so-called bi 1 inear effect at the digital image ° However, the others have the disadvantages of digital image or distortion, the shortcomings of the latter a 1 In view of this, the purpose of the present invention is to propose a uniformity for the above-mentioned image scaling problem. Method for smooth scaling effect 'and its processing

416231 V. Description of the invention (2) The fascination of image zooming is also fast. · The digital image reduction method provided according to the purpose of the present invention described above can be divided into two major steps: First, determine the pixel position of the new image after zooming to obtain an area around the pixels of the new shirt image. The area mapping method is used to map the coordinates of the original pixels of the old image (represented by f in the following) to the pixels of the new image ( The position of the coordinates of the master in the following — & ^ 1 + mountain ^ §). Then it was decided to produce a new image of Xu Yi's smoke work while the mountain stream w < pixel breeding °. For the reduced image work, the part of the old image in the old scene image, for the pixel g of the new image The pixels will be deleted and will not correspond to the new prime: f and its neighbors fail to correspond to the acquisition of the new image material. 'It is obtained by the image of the old image. The data of the pixel points constituting the new image is taken on the flat image These pixels are based on the pixel f as the top SJt 6c pixel g. In the old image, only the dots are enlarged. In the new image, there is a well-known rectangle. As for the image image, but additional pixels are needed, the old pixels will not correspond to the old pixels and will correspond to the new image—the age & v image constitutes a rectangular four-point image with dry people. # W and 丨 respectively) 'The pixel data is taken from ~ " * (assuming that the principle of inverse ratio of horizontal and vertical length distance is similar to the interpolation method. The four pixels are used to calculate any pixel in the new image rectangle—pixel and The detailed process of $ obtaining is: dx and dy are separated, and (w-dx) ^ ^ the horizontal and vertical distances of the rectangle vertices from the four pixels of the image rectangle (H dy) are used as weights to calculate the old value. According to this, a digital image may be a weighted average of the water owners. In the case of zooming in, you can obtain new images through the steps of zooming out or zooming in both horizontally and vertically.

416231 V. Description of the invention (3) In the case of horizontal reduction, vertical enlargement, or horizontal enlargement and vertical reduction ', an intermediate image must be generated first. This intermediate image is also the same as the above-mentioned zooming step, and the old image is first reduced horizontally. / Enlarged ratio, and the vertical enlarged / reduced ratio is 1; the subsequent 1 is similar to the old image for the vertical enlarged / reduced ratio of the old image. At this time, the horizontal direction is reduced / enlarged. The ratio is 1 ', which corresponds to the intermediate image to the new image to be scaled. Of course, in the above, the horizontal or vertical scaling processing order of the image can be reversed. Similarly, the enlargement or reduction processing order can also be reversed. In order to make the above and other objects, features, and advantages of the present invention more comprehensible, a comparative example is given below in conjunction with the preferred embodiments, and the following detailed description is given in conjunction with the accompanying drawings. [Illustration of the figure] Figure 1 shows a partial icon representing each pixel in the screen image with a grid diagram; Figure 2 shows the screen image of Figure 1 reduced horizontally to 40% and vertically reduced to 50% The new image formed; Figure 3 shows the new image formed by horizontally enlarging the screen image of Figure 1 to 320% and vertically zooming to 200%; Figure 4 shows the screen image of Figure 1 at An intermediate image generated by reducing the vertical direction to 60% and horizontally to 160%; Figure 5 shows the screen image of Figure 1 reduced vertically to 60% and horizontally enlarged to 160% , A new image formed after the intermediate image in Figure 4; and

Page 6 416231 Five 'Explanation of the invention (4) Figure 6 shows a flowchart of the steps of the digital image scaling method of the present invention. [Detailed description of the invention] The method of the present invention can be divided into two parts to explain the steps of zooming the screen image: 1. Calculating the mapping of the pixel coordinates of the screen image before and after scaling; and 2. Calculating the pixel data of the new image after scaling: For zooming out , Average the pixels of the old image in the area centered on the pixel coordinates of the new image. For zooming, a series of new pixels are generated based on the inverse distance principle in the area where the coordinates are still the center ^ The following will describe in detail the mapping relationship of the screen image coordinates before and after reduction and enlargement, and the new image in conjunction with "Figures 1 to 5" Algorithm for pixel data. First, the meaning of the labels in the icon will be explained: f (i, j), i = 0,1…, M-1; j = 0, 1…, N -1 represents the pixel data of the original image of M rows and N columns, It is labeled f 1, j in the illustration. With g (x, y), x = 0,1 ‘· ,! ? — 1; y = 0, 1 ....., S-1 represents the pixel data of the new image of Ri 亍 S 歹! J, which is marked as g i, j in the icon. Per represents the scaling value of the original image to the new image. First step of the algorithm: Calculate the relationship between the pixel coordinates of the original image and the new image: X = (i * p r r) integer part y = (j * per) integer part Second step of the algorithm:

416231

416231 V. Description of the invention (6) Step According to the algorithm, the pixel position of the new image (0, 〇) (0, 1) (〇, 2) (1,0) (1, 1) (1,2) (2 , 0) (2,1) (2, 2) Pixel position of original image (0, 0) (0, 3) (0, 5) (2,0) (2,3) (2, 5) (4 , 0) (4, 3) (4, 5) and so on. According to the second step of the algorithm:), with f (0, 0) as vertices, f (0, 2), f (1, 0), for g (in Figures 1 and 2) in this rectangle ( 0, 〇) corresponds to f (〇, plus f (0,1) ί (1, 1), f (1, 2), which does not appear in the new screen image. Pixel g (0,0). That is g (0, 0) = (f (0,0) + f (0, 1) + f (Ο, 2) + f (1, 0) + f (1,1 ) + f (1,2)) / 6 Similarly, g (0, l) = (f (0, 3) + f (0, 4) + f (1, 3) + f (1,4)) / 4 g (1,0) = (f (2,0) + f (2,1) + f (2,2) + f (3,0) + f (3, 1) + f (3, 2) ) / 6

416231 V. Description of the invention (Ό g (l, l) = (f (2, 3) + f (2, 4) + f (3, 3) + f (3, 4)) / 4 etc. The actual horizontal and vertical magnification will be explained with reference to actual operation. Please refer to "Figures 1 and 3". The original image of "Figure 1" is magnified horizontally to 3 2 0% and vertically to After 2 0 0%, "the graph follows the first step of the algorithm: the pixel position of the new image (0, 0) (0,3) (0,6) (2,0) (2, 3) (2, 6) (4,0) (4,3) (4,6) Pixel position of original image (0, 0) (0, 1) (0, 2) (1,0) (1,1) (1,2) (2, 0) (2,1) (2, 2), etc. According to the second step of the algorithm: f (0, 0), f (0,) in the old image, and g (0, 0) rectangle with vertices, in this calculation 2), g (l, 〇), g (l, l), g (l, 1), f (1,0), f (1,1) four pixels , G (0, 3), g (2, 0), g (2, 3) are six pixels that will produce g (0,0), g (0, l), and g (0, 2). Among them W = 2,

Page 10 416231 V. Description of the invention ¢ 8) H = 3, now calculate g (l, 1) pixels, dx = l, dy = l with g (〇, 0) (corresponding to f (0, 0)) With dx = l, dy = 2 for g (0,3) (corresponding to f (0, l)), and dx = l, dy = l for g (2, 0) (corresponding to K1, 0), and g (2, 3) (corresponding to f (1, 1)) dx two 1, dy = 2 'thus: g (i, 1) :( f (0, 0) * ((2-1) + (3 -1)) + f (0,1) ^ ((2-1) + (3-2)) + f (l, 0) ^ ((2-1) + (3 ^ 1)) +

Kl, 1) wood ((2-1) + (3-2))) / (((2-1Η (3-1) Η ((2-1) + (3-2) η ((2-1 ) + (3-1)) + ((2-1) + (3-2))) Similarly, g (Ο, Ο) = (f (Ο, Ο) wood ((2-Ο Η (3-Ο )) + f (Ο, 1) * ((2-Ο) + (3-3)) + f (1, 0) ^ ((3-0) + (2-2)) + f (l, 1 ) ^ ((2-2) + (3 ^ 3))) / (((2 ^ 0) + (3-0)) + ((2-0) + (3-3)) + ((3- 0) + (2-2)) + ((2-2) + (3-3))) g (0, 1) = (f (0, 0) ^ ((2-0) + (3-1 )) + K〇, 1) * ((2-0) + (3-2)) + f (1, 0) * ((2-2) + (3-l)) + f (l, 1) ^ ((2-2) + (3-2))) / (((2-0) + (3-1)) + ((2-0) + (3-2)) + ((2-2 ) + (3-1)) + ((2-2) + (3-2))) g (0, 2) = (f (0, 0) ^ ((2-0) 4- (3-2 )) +

Page il 416231 V. Description of the invention (9) f (0, 1) * ((2-0H (3-1)) + f (l, 0) * ((2-2) + (3-2)) + f (l, 1) * ((2-2) + (3-1))) / (((2-0) + (3-2)) + ((2-0) + (3-1) ) + ((2-2) + (3-2)) + ((2-2) + (3-1))) g (l, 0) = (f (〇, 0) ^ ((2-1 ) + (3-0)) + f (0,1) ^ ((2-1) + (3-3)) + f (l, 0) ^ ((2-1) + (3-0)) + f (l, 1) wood ((2-1) + (3-3))) / (((2 ^ 1) + (3-0)) + ((2-1) + (3-3) ) + ((2-1) ten (3-0)) K (2-1) + (3-3))) g (l, 2) two (f (0, 0) * ((2-1) + (3-2)) + f (0,1) ^ ((2-1) + (3-1)) + f (l, 0) ^ ((2-1) + (3-2)) + f (l, 1) * ((2-1) + (3-1))) /. (((2-1) + (3-2)) + ((2-1) + (3-1) ) + ((2-1) + (3-2)) + ((2-1) + (3-1))) and so on. In the following, the horizontal and vertical directions will zoom in and out each other, and the actual operation will be explained. Please refer to "Figures 1 and 4, and 5" again. The original image of "Figure 1" is reduced to 60% in the vertical direction and enlarged to 160% in the horizontal direction to form Figure 5 "" Figure 4 "is generated Intermediate image. First, the original image is reduced to 50% in the vertical direction.

Page 12 416231 V. Description of the invention αο) The horizontal direction is reduced to 100% t (x ,, y '). According to the first step of the algorithm, the pixel position of the original image is (0, 0) (0, 1) (0, 2) (2, 0) (2, 1). 0, 0) (0,1) (0, 2) (1,0) (2,2) (1,2) (4, 0) (2, 0) (4, 1) (2, 1) ( 4, 2) (2, 2) and so on. According to the second step, the mean value becomes several points in one dimension and averaged: t (0, 0) = (f (0, 0) + f (1, 0)) / 2 t (0, 1) = (f (0, l) + f (l, 1)) / 2 t (l, 0) = (f (2, 0) + f (3, 0)) / 2 t (l, 1) = (f (2, l) + f (3, 1)) / 2 and so on. Then, the intermediate image is enlarged to 16 ％ in the horizontal direction, and then enlarged to 1 ◦ 0% in the vertical direction. The new image g (X, y) of “Figure 5” is obtained. According to the first step of the algorithm:

416231 g (〇, 1) t (l, 2) * ((1-m (2-((1-0) + (2-0)) + ((1 ((1-1) + ( 2-0)) + ((1 ((1 ^ 0) + (1-0)) D) 〇)) -0)) + ((1 -0)) + ((1 1-0) + (2 -1-0) + (2-1-1) + (2- V. Description of the invention (11) Pixel position of the intermediate image (0, 0) (0, 1) (0, 2) (1,0) ( 1,1) (1,2) (2,0) (2, 1) (2, 2) etc. Follow the second step of the algorithm: g (0, 0) = (t (0, 0 (0, 1) * (((1, 〇) * (((1,1) wood (((1-0) + (1 ((1-1) + (1 t (0,1) ^ ((t ( 0,2) * ((t (l,!) * ((Pixel position of new image (0, 0) (0, 1) (0, 3) (1,0) (1,1) (1, 3 ) (2, 0) (2, 1) (2, 3) 1-0) + (1-1 ^ 1) + (1-1-1) + (1-1))) / -0) + ( 1-1)) + -1) + (1-1))) 0)) + 2)) + 〇)) + 2))) / -0) + (2-2)) + -1) + ( 2-2))))

Page 14 V. Description of the invention (12) g (〇, 2)-(t (0,1) ^ ((1-0) + (2-1)) + t (0,2) ^ ((1- 0) + (2-1)) + t (l, 1) * ((1-1) + (2-1)) + t (l, 2) * (Um (2-l))) / (( (1-0) + (2-1)) + ((1-0) + (2-1)) + ((1-1) + (2-1)) + ((1-1) + (2 -1))) g (l-0) " (t (l, 〇) ^ ((l-〇) + (l-〇)) + t (l, 1) * ((1-0) + ( 1-1)) + t (2, 〇) * ((ii) + (i-〇)) + t (2,1) 氺 ((: l -1) + (: i-i))) / ( (〇-0) Kl-0)) + ((l-0) Kl-1)) + ((1-1) + (1-0)) + ((1-1) + (1-1)) ) And so on. Finally, the process of reducing the digital image according to the present invention will be described with reference to "Fig. 6". = T 1: At the beginning of the program, first determine whether the image is to be enlarged horizontally? Whether it is horizontally enlarged or reduced, then continue to determine whether it is vertically enlarged. Steps 2 and 3: Determine whether it is vertically enlarged or reduced. Step 4: If it is “horizontal magnification and vertical magnification”, the pixel position mapping of water magnification and vertical magnification is calculated. Large step 5 'realizes horizontal and vertical magnification. Step 61 1 is to zoom in horizontally and zoom out vertically, then the intermediate image can be generated first. Ding ^ + put

Page 丨 416231 V. Description of the invention (13) Step 7: Then, perform vertical reduction based on the intermediate image to generate pixels of a new image. Step 8: If it is zoomed out horizontally and vertically, it can be zoomed up first to generate an intermediate image. Step 9: Then reduce horizontally on the basis of the intermediate image to generate pixels for the new image. Step 10: If it is horizontal reduction and vertical reduction, calculate the pixel position mapping of horizontal reduction and vertical reduction. Step 11 1. Reach the horizontal and vertical reduction. The enlargement and reduction algorithms are as described above. Although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some modifications and retouching without departing from the spirit and scope of the present invention. In the method of the invention, the pixel coordinate of the new image is an integer portion obtained by multiplying the pixel coordinate of the original image by the scaling value. Of course, the pixel coordinates of the new image can also be calculated by rounding or unconditional rounding; in addition, the mean value at the time of reduction Calculated pixel selection can also use other selection methods; the same is true when zooming in; furthermore, if horizontal zoom vertical zoom or horizontal zoom vertical zoom image scaling is processed, the horizontal or vertical zoom processing order of the image It can be reversed, and the processing order of enlargement or reduction can also be reversed. Therefore, the scope of protection of the present invention shall be determined by the scope of the attached patent application.

Page 16

Claims (1)

mm 6. The scope of patent application [Scaling method of digital image to achieve ― horizontal and imaginary image of a scene and image +, ^ The sub-style screen takes P% π and the direction of the straight direction is the same as% • zoom out operation, It includes taking a plurality of bases and brackets of the image, reducing the ratio of the horizontal and vertical directions by two, respectively, and shooting the position of the quasi-pixel to the system—new ^ ζ the base 2 / mean 1 to obtain Corresponding to the new pixel; a new image formed by the new pixel is created with & , :: Please patent No.! The digital image mentioned in item 1 is placed as follows: Ϊ: ί: Set the corresponding reference pixel position: take the value of position M and take the integer part as the new pixel = digital image scaling described in item 1 Method 'MT: The new pixel position is obtained by multiplying the position of the reference pixel by its left coordinate and then setting the coordinates. Pick up the new pixel bit 4 'as in the first item of the scope of patent application f where the mean calculation includes the base m scaling :: method' The reference pixel of the pixel is an item; it is composed to correspond to the new 'a kind of digital Image zoom method, 5. Water: and vertical magnification at the same time:: Now = the screen image is taken to capture the reference pixels of the image;, the position of the element is mapped to the value of Μ $ Μ π respectively, and the reference image is mapped to the The position of the new pixel; On page 17, six fields are placed in the horizontal and vertical directions respectively. 7 8 416831 The scope of the patent application obtained four points, such as # Occupies the new pixel area corresponding to the adjacent reference + pixel field. Add the new pixel area and make the unmapped reference pixels 均 + average to obtain the pending pixel; to create a new image formed by the new I $, such as the pixel in the scope of the patent application. The new pixel position is obtained by multiplying the position of the reference pixel position by multiplying the position of the reference pixel position by multiplying the two-dimensional image scaling method. Take the integer part of the X pixel value as the new pixel, as described in the patent application. The new pixel position is: the method of digital image scaling. Its position coordinate is multiplied by the reference pixel position that the weight should be. , Round the coordinates to the ratio value as the new pixel position = the 5th area of the patent range where the new pixel area has a length and width of each method, and the four-point reference pixel distance is (dx, dy). ) κη,) are weights, and the four points are weighted averages. A thousand lanes are not known: a method of digital image scaling to achieve horizontal reduction and vertical enlargement. Ϊ́ :: like the horizontal reduction-the first-ratio = vertical ..., part of the benchmark Like I # n 钿, the position of a small prime is mapped to the corresponding one — page 18, 416,231, six. The position of one pixel of the patent application range; any-the reference pixel and its adjacent unmapped prime are averaged to obtain The corresponding reference image of the pixel is created with the first pixel—the intermediate image. ',, enlarged in the vertical direction—the second ratio value, the level is 1, the small pixel i γ 15 fine ratio The position of the two pixels; the position maps to the corresponding—the fourth area that is adjacent to the fourth—the image-like F area; # ^ the corresponding second pixel is equal to four points of the first pixel to the second pixel area = -To-be-taken pixels are weighted average 'to obtain the to-be-determined J prime ... like a new image formed by 10 pixels and the to-be-determined pixel. I 0 'If the scope of patent application is q 顼 丨 豕 豕 可 The process of reducing and enlarging in the vertical direction is possible II Please? The digital image zooming method described in item 9 of the range of interest ^ ί ί One pixel position or the second pixel position is the corresponding Υ > pixel position or the first pixel position, and its position coordinates are multiplied by f Take the integer part after the proportional value as the coordinate of the first pixel position or the second pixel. , I :: The method of digital image zooming described in item 9 of the monthly patent encirclement :: ° The first pixel position or the second pixel position is the corresponding / local pixel position or the first pixel position, and its position Coordinate multiplication Scope of patent application The above mentioned Hai-Hai comparison value is rounded to be the first pixel position or the second image 3, and the coordinates. Among them, the method of digital image scaling described in item 9 of the second range is the first pixel Xin Zhi ^ f ^, the reference pixel system constitutes an area corresponding to the 4 pixels as a vertex. τ Ming patent scope ninth, where the length of the second pixel area == the method of image scaling is different from the four-point first-pixel distance ❹ and Η, the pending pixel score is a weighted average calculated by +, weighted average . Value, find the four-th pixel-page 20