RU2469400C1 - Method to convert bitmapped image into metafile - Google Patents

Abstract

FIELD: printing industry.

SUBSTANCE: background area is detected in a bitmapped image; a background type is detected; a command is saved into a metafile, which relates to background display; multicoloured areas are detected in a bitmapped image; multicoloured areas are saved into a metafile as a command related to display of bitmapped image fragments; single-coloured areas are detected in a bitmapped image; a command is saved into a metafile related to display of single-coloured areas.

EFFECT: provision of high quality of display of a bitmapped image converted into a metafile, with considerable reduction of saved data volume compared to memory volume required to store an initial digital bitmapped image.

7 cl, 17 dwg

Description

The invention relates to digital image processing technologies, and more particularly, to methods for converting raster images to electronic format.

Modern methods of converting raster images into an electronic format convenient for storing and reproducing them are faced with the need to solve two often conflicting tasks. The first task is the need to save information in a format that requires as little memory as possible. The second task is to prevent degradation (significant deterioration) of the original image as a result of its conversion to electronic format. One of the most common methods for such image conversion is Mixed Raster Content (MRC) technology (see, for example, OSO / IES JTC1 / SC29 / WG1 N542, "ITU-T Mixed Raster Content model as JPEG 2000 Architectural Framework," June 30, 1997 ) [1], which allows reducing the size of the resulting file through the use of a multilayer image representation model and various compression methods oriented to compressing each layer in the most suitable way for this purpose. However, MRC algorithms do not take into account the features of the display of vector graphics present on the original bitmap image. If, for example, in the original electronic text document, the display and printing of characters does not depend on their scale, then the display of the characters of the scanned text document converted by MRC will depend on the scale of their reproduction. It is also worth noting that the well-known methods of multilayer analysis and preservation of raster images in electronic form separate all raster images into a separate layer. Thus, even if only minor elements of the raster graphics are present on the original image, a raster layer with a size comparable to the size of the entire document will be saved in the resulting document.

In the prior art, in particular, a method for reducing file size disclosed in US patent application No. 20110007334 [2] is known. This technical solution proposes a method and system for vectorizing text on a scanned image. The authors disclose a vectorization method that provides a smooth display of characters. The disadvantage of this solution is that only the test present on the scanned image is analyzed, and accordingly, the vectorization procedure is considered independently of the rest of the image elements.

Closest to the claimed invention features has the method described in US patent No. 7289122 [3]. The method is based on the application of approximation of the shape of graphic objects using Bezier curves. The disadvantage of this solution is that the authors disclose only the approximation method, while the conversion of the entire image to the form of an electronic document is not considered.

The problem to which the invention is directed, is to develop an improved method for converting a raster image to a metafile. The inventive method should provide improved display quality of the bitmap converted to a metafile, while significantly reducing the amount of data stored in comparison with the amount of memory required to store the original digital bitmap.

The technical result is achieved through the application of a method of converting a raster image to a metafile, which includes the following operations:

- reveal the background area on the raster image;

- determine the type of background;

- save the commands related to the display of the background in the metafile;

- multicolor areas are detected on the raster image;

- save multi-color areas in the metafile as commands relating to the display of fragments of the bitmap image;

- single-color areas are detected on the raster image;

- save the commands related to the display of monochrome areas into the metafile.

The resulting metafile combines both raster and vector descriptions of visual information. Effective conversion of the original raster image into a metafile, from the point of view of the ratio of display quality to the amount of stored data, is achieved by combining the most suitable methods for converting and storing data corresponding to specific categories of areas in the image. For example, fragments of bitmap graphics present in the original image and including illustrations, photographs, drawings, are localized and stored independently in the metafile. Vector graphics elements, such as lines, tables, text, logos, etc. they are detected and converted into a vector format by describing their shape and color with the metafile graphic commands.

Thus, the claimed method combines two main stages. The first step is to identify areas of predefined categories in the original bitmap. These categories include: background images, multicolor and monochrome areas. The second stage includes an effective description of each of the categories of areas with the corresponding metafile commands and their preservation. The background of the image is described by the metafile commands depending on its type. For example, a single-color background is set by a command that determines its color, a multi-color background is a background bitmap image whose size corresponds to the size of the displayed document. Multicolor local areas are saved in the metafile as fragments of raster images, the location and size of which are set by the corresponding metafile commands. In addition, the shape of non-rectangular multicolor regions is defined by the corresponding clipping regions described by the graphic metafile commands. Monochrome areas of the image are saved to the metafile in vector form using the corresponding graphic metafile commands.

Further, the essence of the claimed invention is illustrated with the use of graphic materials.

Figure 1 A block diagram illustrating the principle of converting a raster image to a metafile.

Figure 2 An example of a bitmap with various types of areas.

Figure 3 An example of overlapping areas during the display of the metafile.

Figure 4 Illustration of a system for converting raster images to metafile.

5 A flowchart of a process for storing a multicolor region in a metafile.

6 A flowchart of a process for storing a single-color region in a metafile.

Fig. 7 An example of determining the external contour of a region.

Fig. 8 An illustration of a technique for determining segments of prospective key contour points.

Fig.9 Illustration of the process of determining the key points of the contours on the example of a single-color region.

Figure 10 An example of simplified contours of a single-color region.

11 Block diagram of the approximation process of the simplified contour of the region by a sequence of line segments and curves.

Fig. 12 Example of adjustment of coordinates of key points of a simplified contour.

Fig. 13 Example of Bezier curves.

Fig. 14 Illustration of a process for calculating control points of a Bezier curve.

Fig. 15 Illustration of the approximation of a Bezier curve.

Fig. 16 Example of approximation of the contours of a single-color region by a sequence of Bezier curves.

Fig. 17 Examples of the original monochrome areas and the results of their display by playing the metafile.

For an unambiguous interpretation of the terms used hereinafter, it should be clarified that a metafile usually refers to the format of such files that store data of various types, in particular raster and vector data types. Typically, a metafile structure is a sequence (list) of commands (metafile records) with a possible set of arguments. When executing (playing) a metafile, each command performs a specific action, for example, rendering (drawing) a line segment, displaying a fragment of a raster image, etc. As a non-limiting example, we can mention metafiles represented in PDF, XPS, PS, EMF, etc.

The key steps of converting a bitmap image to a metafile are illustrated in FIG. The original bitmap 101 may be obtained from any device suitable for capturing or receiving a bitmap. At step 102, identifying areas of the background image and determining the type of background is performed. In a preferred embodiment of the inventive method, the background is image areas forming a background and correlated with one of three predefined categories: a single-color background, a gradient background, and a multi-color background. The monochrome background includes, in particular, background areas of the image characterized by approximately the same level of brightness and color. The gradient background includes, in particular, image areas with a uniform change in brightness and color, moreover, such a change is described by a specific function. A multicolor background includes background areas of a raster image that cannot be assigned to a monochrome or gradient background. For example, a background bitmap, including texture, can be classified as a multicolor background. The detected background is described using metalanguage commands corresponding to the selected metafile format. As a non-limiting example, we can mention the possible metafile commands for each background category. A single-color background is defined by metafile commands that indicate the background color. For a gradient background, the metafile commands determine the type of gradient transformation, the direction of the gradient drop, and the range of brightness or color. A multicolor background corresponds to a background bitmap saved in a metafile. Depending on the embodiment of the inventive method, the background may be replaced with a transparent one that is not displayed during the execution (playback) of the metafile. A similar option can be implemented in the form of settings of a device or system that implements the claimed method. At step 104, multi-color areas containing raster images or complex graphics are detected. In this case, a multicolor region refers to a group of connected pixels (dots) of a raster image localized in some part of the original raster image. Detected multicolor regions are stored in the metafile 105 as bitmap images, followed by the corresponding metafile commands describing the parameters of each multicolor region and their position in the image. At step 106, connected monochromatic areas, that is, areas with approximately the same brightness and color, are detected. At this stage, monochrome areas are described by a set of graphic metafile commands that implement the display of vector graphics elements approximating the shape of these areas. The result of the vector description of monochrome areas is stored in the metafile at step 107. In a preferred embodiment of the proposed method, compression technologies are used to reduce the size of the metafile, while metafile records are compressed without loss, and compression with loss of quality is allowed for bitmaps stored in the metafile.

The inventive method provides an effective approach to converting raster images to a metafile in terms of the ratio of display quality to file size. This is achieved by storing each type of visual information in the most appropriate way. In particular, the vector description of monochrome areas ensures that their display is independent of the visualization scale or, for example, the print scale. For ordinary bitmap images, zooming in causes degradation of their display.

Figure 2 illustrates an example of a bitmap image. The figure shows the main categories of regions described in the claimed method: background 201, monochrome areas 202 and 205, rectangular multicolor region 203 and non-rectangular multicolor region 204.

Figure 3 illustrates the visualization order of each category of areas when playing (playing) a metafile. In this case, the order of preservation of the areas illustrated in FIG. 1 corresponds to the order of their visualization. As an example, we can clarify that the visualization of metafiles often occurs by superimposing on each other transparent or translucent layers or areas. In the process of displaying the metafile, background 303 is the bottom layer onto which the multi-color areas 302 and then the single-color areas 301 are superimposed.

Figure 4 schematically illustrates a computing system that implements the inventive method. A computing system that converts raster images to a metafile includes: a device 402 for inputting an initial raster image, a processor 403 and a memory 404 that stores instructions of a program 405 implementing the inventive method. Data transfer between the system modules is carried out via the data bus 401. As a non-limiting example of the implementation of the proposed method, the input device 402 can be: a scanner, a digital camera, a memory buffer that stores a screenshot (screenshot), etc. The metafile obtained as a result of executing the instructions of the program is stored in memory, from where it can later be transferred, for example, to a printing device, visualization device, recorded in a data storage device, etc. Only those features that are mentioned in the description are illustrated. However, it should be understood that the computing system may have additional features that were not reflected in the illustration. The computing system can be represented, for example, by a personal computer, mobile phone, television, multifunction printing device, or any other electronic device.

Figure 5 schematically illustrates the main steps that ensure in the aggregate the preservation in the metafile of each identified multicolor region. At step 501, the bounding rectangle of the analyzed multicolor region is determined. The bounding box describes the pixels of the multicolor region in accordance with its position in the original image, including the definition of the bevel angle of the region. This allows you to reduce the size of the image stored in the metafile. Step 502 aims to determine the shape of the multicolor region. In a preferred embodiment of the proposed method, the multicolor region is rectangular or non-rectangular (arbitrary) in shape. A more accurate display of a non-rectangular multi-color region is achieved by defining its clipping region in step 503. The need for this step is due to the fact that only a rectangular image can be saved in the metafile. The clipping region allows you to hide the non-displayed portions of the saved fragment of the raster image for a non-rectangular multicolor region. This step will be described in more detail below. Step 504 zooms in on a portion of the input bitmap. The location of the fragment is determined by the coordinates of the bounding box.

Scaling is usually performed in the direction of reducing the size of the original image to a predetermined size, preferred for subsequent playback of the metafile. For example, the original bitmap scanned at a resolution of 300 dpi (dots per inch) can be reduced to a resolution of 72 or 96 dpi, corresponding to the resolution of the display. At step 505, the shape of the multi-color region is described by using metafile commands corresponding to the selected format. For example, for a rectangular multicolor region of a metafile command, the transform matrix, the size of the saved fragment of the image, the method of its compression, color space, etc. are usually determined. The transforming matrix reflects a number of parameters that affect the display of a fragment of a raster image when playing a metafile. As an example, these parameters include: resolution of the image fragment rendering (this resolution may differ from the resolutions of the original bitmap image and the image fragment stored in the metafile); displacement of the image fragment relative to the coordinates of the visualized document as a result of playing the metafile; angle of rotation. For a non-rectangular area, metafile commands also include a series of graphical metafile commands that define the shape of the clipping region. At step 506, the metafile commands and the fragment of the saved raster image are compressed. Compression is necessary to reduce the size of the resulting metafile, but, nevertheless, is not a mandatory step, and depending on the implementation of the claimed method, this step may be excluded. The compression methods used in step 506 should be supported by the selected metafile format. For compression of metafile commands, compression methods without loss of quality are used, for a fragment of the image, methods of compression of bitmap images with loss of quality are acceptable. Step 507 stores the compressed metafile instructions and the fragment of the bitmap image corresponding to the multicolor region into the metafile.

6 illustrates a simplified block diagram of storing each single-color region in a metafile. As mentioned above, monochrome areas are preserved by describing the shape and color of each area with a set of graphical metafile commands. Such a method of converting a raster region of an image into a vector provides high quality display of the region and independence of the viewing scale. At step 601, tracking of each contour of the analyzed one-color region, including external and internal contours, is carried out. An area can be limited to only one external contour. There may be several internal circuits, or they may be absent. At 602, the fill color of the monochrome region is determined. Next, at step 603, the contour points are approximated by sequences of straight line segments and curves supported by the selected metafile format. The sequences of the approximating segments and the fill color of the one-color region are described by the corresponding metafile commands in step 604, which are then compressed (step 605). Step 606 saves the compressed instructions to a metafile. The clipping region of non-rectangular multicolor regions, determined in step 503, is calculated in a similar manner, except that only the outer contour of the region is approximated and there is no step for determining the color of the region. For a non-rectangular multicolor region, the approximating sequence is indicated in the metafile as the clipping region.

7 illustrates an example of tracking (tracing) the contour of an area. After the procedures for identifying areas in the raster image, the elements of each area are separated from the background and are related groups of pixels (points). In the conventional representation of bitmap images, the location of the pixels is determined by integer coordinates and the pixels themselves are square. In a preferred embodiment of the inventive method, the contour line is defined as the boundary between the pixels of the background and the analyzed area. In accordance with the illustration of Fig. 7, the pixels of the region are indicated by the dark gray squares 701. Accordingly, the contour passes through the points (vertices) located at the corners of the pixels and marked with white dots 702. These contour points have so-called "virtual", non-integer coordinates. The loop tracking (tracing) procedure starts from some starting point of the contour 703 and continues along the contour in a predetermined direction until the starting point meets again. The direction of tracking the contour is determined by the requirements of the selected metafile format, for example, for the non-zero winding rule, the external contour is tracked (traced) in the clockwise direction, and the internal contours are counterclockwise or vice versa. Such a rule ensures the correct filling of the area with the specified color during the playback of the metafile.

After tracking the contours of a single-color region, the number of contour elements is reduced (contour simplification) by determining their most significant key points. The procedure for finding the key points of the contour includes two main steps: at the first step, determine the segments of the contour, within which the estimated key points are located; in the second step, from the possible combinations of key points, choose a combination that provides the smallest approximation error. FIG. 8 illustrates an example of calculating a contour segment within which an estimated key point is located. To do this, each contour point is associated with a pair of the most distant points in accordance with a predetermined criterion. For example, a contour point 803 corresponds to a pair of the most distant points 802 and 806. In a preferred embodiment of the proposed method, two contour points are assumed to be the most distant from each other if the portion of the contour between them can be approximated by a straight line, and the approximation error does not exceed a predetermined value. The approximation error is calculated as the sum of the squares of the distances 804 from each point of the approximated section of the contour 805 to the corresponding approximating line 801. The set of pairs of the most distant points allows you to determine the sections of the contour that are preferable for determining key points. For example, the point 809 in Fig. 8 is the farthest for the group of contour points indicated by black dots 807 in the illustration, respectively, the point 808 is the farthest for all contour points indicated in gray 810. Thus, the points 808 and 809 for the analyzed fragment of the contour occur the greatest number of times and, accordingly, determine the boundaries of the portion of the contour 811, preferred for the detection of a key point.

Figure 9 shows an example of a one-color region with certain contour sections preferred for detecting key points. Sites are indicated by shaded areas 902. Only one key point can be detected within a single site. The following are possible combinations of key points within the respective sections. Among them, a combination is selected that corresponds to the minimum error of approximation of the contour by segments of straight lines. The illustration of Fig. 9 shows the resulting key points 901. Their corresponding approximating combination of line segments with a minimum approximation error is shown in Fig. 10.

Based on the key points of the contour, the procedure of approximating the contour is performed by a sequence of line segments and curves. Schematically, this procedure is illustrated in the flowchart of FIG. 11. Step 1101 performs adjustment (refinement) of the coordinates of the key points of the simplified contour of the area. As shown in FIG. 12, new approximate segments 1203 correspond to new key points 1201 with refined coordinates, providing a smaller error in the approximation of the contour. Moreover, the coordinate adjustment is limited to the vicinity of the source key points 1202. In a preferred embodiment of the claimed method, said neighborhood corresponds to the size of one pixel, i.e. the distance between the specified and the original location of the key point does not exceed half a pixel. Step 1102 approximates a simplified outline defined by refined key points, line segments, and curves corresponding to the selected metafile format. At this stage, the number of vector graphics elements used is uniquely determined by the number of key points. Step 1103 performs a reduction in the number of curves involved in the approximation, which provides a smoother display of the contours during playback of the metafile and a smaller number of required graphic metafile commands. At 1104, sequences of line segments and curves are converted (converted) into metafile instructions describing them.

In a preferred embodiment of the proposed method, the approximation of the contours of the region is performed using two graphical functions supported by most existing metafile formats: a straight line segment and a third-order Bezier curve. The command to display a straight line segment is usually accompanied by arguments defining the beginning and end of the line segment. Accordingly, the third-order Bezier curve visualization command includes arguments defining the beginning, end of the curve, and two control points that control the bending and shape of the curve.

Bezier curves are used to describe the contour sections formed by the intersection of two approximating segments at the key point of the contour. Thus, when calculating the parameters of the curve, a triangle is analyzed, one of the vertices of which is the key point, and the edges adjacent to it are approximated segments. In accordance with the illustration of FIG. 13, to simplify the calculation of the parameters of the Bezier curve, it is assumed that the control points 1302 of the curve are located on the faces of the analyzed triangle adjacent to the vertex 1301. In this case, the control points of the curve are determined by two parameters α and β, whose changes in the range from 0 to 1 are illustrated by several examples in FIG. 13. In the general case, the values of the indicated parameters may exceed unity.

. Полученная путем параллельного переноса прямая 1406 пересекается с аппроксимирующими отрезками в точках 1402 и 1405, определяющих контрольные точки кривой Безье α и β соответственно. Значения α и β с точностью до пропорционального коэффициента равняются нормированным длинам отрезков, отсекаемых параллельными прямыми. На примере, приведенном на Фиг.14, параметр α равен отношению длины отрезка, ограниченного точками 1401 и 1402, к длине отрезка 1401-1403, умноженному на пропорциональный коэффициент. Предпочтительное значение пропорционального коэффициента равняется 4/3, в этом случае кривая Безье проходит через точку пересечения с окрестностью ключевой точки (1409, 1507). На Фиг.15 приводится кривая Безье, соответствующая вычисленным параметрам. Аппроксимирующая кривая 1505 Безье соединяет средние точки 1501 и 1506. Контрольные точки 1502, 1504 кривой вычислены в соответствии с приведенным выше описанием.Fig and Fig illustrate the steps of calculating the control points of the Bezier curve. As a first step, the midpoints 1401 and 1407 of the corresponding approximating segments of the simplified contour intersecting at the analyzed key point 1403 of the contour are calculated. Next, line 1408 passes through the indicated midpoints. At the next step, the resulting line is parallel transferred until it intersects at point 1409 with the vicinity of the key point. The neighborhood of the key point is bounded by a circle 1404 of a predetermined radius. In a preferred embodiment, the radius value corresponds to the length of the pixel side of the bitmap multiplied by the coefficient

. The straight line 1406 obtained by parallel transfer intersects the approximating segments at points 1402 and 1405 that define the control points of the Bezier curve α and β, respectively. The values of α and β are accurate to a proportional coefficient equal to the normalized lengths of segments cut off by parallel lines. In the example shown in Fig. 14, the parameter α is equal to the ratio of the length of the segment bounded by points 1401 and 1402 to the length of the segment 1401-1403 times the proportional coefficient. The preferred value of the proportional coefficient is 4/3, in which case the Bezier curve passes through the intersection point with the vicinity of the key point (1409, 1507). On Fig shows the Bezier curve corresponding to the calculated parameters. An approximating Bezier curve 1505 connects the midpoints 1501 and 1506. The control points 1502, 1504 of the curve are calculated in accordance with the above description.

. В случае если параметр γ удовлетворяет диапазону допустимых значений 0,55<γ≤1, параметры α и β остаются без изменений. При выполнении условия γ≤0,55, контрольные точки кривой Безье заменяются значениями, соответствующими α=β=0,55. Если выполняется условие γ>1, тогда принимается решение о замене аппроксимирующей кривой двумя отрезками прямых, ограниченных соответствующей средней точкой и примыкающих одна к другой в анализируемой ключевой точке. Например, выполнение условия γ>1 для Фиг.15 приведет к тому, что кривая Безье будет заменена двумя отрезками. Первый отрезок ограничен точками 1501 и 1503, второй отрезок ограничен точками 1503 и 1506.After calculating the parameters of the Bezier curve, it checks whether its control points correspond to a number of conditions. For this, the parameter is calculated

. If the parameter γ satisfies the range of admissible values of 0.55 <γ≤1, the parameters α and β remain unchanged. When the condition γ≤0.55 is fulfilled, the control points of the Bezier curve are replaced by values corresponding to α = β = 0.55. If the condition γ> 1 is satisfied, then a decision is made to replace the approximating curve by two segments of lines bounded by the corresponding midpoint and adjacent to one another at the analyzed key point. For example, the fulfillment of the condition γ> 1 for Fig. 15 will cause the Bezier curve to be replaced by two segments. The first segment is limited by points 1501 and 1503, the second segment is limited by points 1503 and 1506.

On Fig illustrates an example of approximation of the contours of a single-color region in accordance with the claimed method.

17 illustrates examples of enlarged original raster monochrome areas and the result of their display after approximation. It should be noted that since the proposed approximation corresponds to the vector description of the shape of the regions, therefore, their display does not depend on the viewing scale.

The inventive method is intended for implementation by software and hardware in black and white and color multifunction printing devices and digital copiers. Also, the method can be implemented as part of the software scanning devices.

Claims (7)

1. A method of converting a raster image to a metafile, including the following operations:
- reveal the background area on the raster image;
- determine the type of background;
- save the commands related to the background display in the metafile;
- multicolor areas are detected on the raster image;
- save multi-color areas in the metafile as commands relating to the display of fragments of the bitmap image;
- single-color areas are detected on the raster image;
- save the commands related to the display of monochrome areas into the metafile. 2. The method according to claim 1, characterized in that the type of background on the bitmap image is determined in accordance with the following predefined categories: single-color background, gradient background, multi-color background. 3. The method according to claim 1, characterized in that the instructions for displaying the background are stored in the metafile in accordance with the background type previously defined:
- one-color background is described as a rectangle filled with the specified color;
- gradient background is described by records in the metafile that determine the direction and type of gradient color change;
- the multicolor background is saved to the metafile as a bitmap image. 4. The method according to claim 1, characterized in that during the determination of the type of background, depending on the preset settings, the background can be replaced with fully or partially transparent. 5. The method according to claim 1, characterized in that the multi-color areas are saved in a metafile by performing the following operations for each multi-color area:
- define the bounding box that describes the multicolor region, and it is allowed that the sides of the bounding box were tilted to the sides of the original bitmap;
- determine the shape of the multicolor region as rectangular or non-rectangular;
- determine the clipping region for a non-rectangular multicolor region;
- scale the fragment of the raster image corresponding to the multicolor region, to a predetermined resolution;
- describe the shape of the multicolor region by means of the corresponding metafile commands;
- compress the metafile commands and the bitmap fragment corresponding to the multicolor region;
- save compressed commands to a metafile. 6. The method according to claim 1, characterized in that the one-color areas are saved in a metafile by performing the following operations for each one-color area:
- track the points of the external and internal contours of the same-color region;
- evaluate the color of the monochrome area;
- approximate the points of the external and internal contours by sequences of line segments and curves;
- form a sequence of metafile commands describing the approximating sequence and color of a single-color region;
- compress the sequence of metafile commands;
- save compressed commands to a metafile. 7. The method according to claim 1, characterized in that when saving the multi-color areas in the metafile, the clipping region for the non-rectangular multi-color area is determined by performing the following steps:
- track the points of the outer contour of a non-rectangular multicolor region;
- approximate the points of the outer contour by a sequence of line segments and curves;
- make up a sequence of metafile commands describing the approximating sequence and defining it as a clipping region for a multicolor region;
- compress the sequence of metafile commands;
- save compressed commands to a metafile.

Images