TW464830B - Approaching process method of digitized vector font - Google Patents

Abstract

The invention relates to an approaching process method of digitized vector font, which is used to process a set of line segments obtained from the conversion of font model and makes it become the set of digitized vector approaching to line segment with terrific look and compact data size. The invention determines the best end points of a vector line segment by means of the approaching tolerance of each line segment node for constructing any vector line segment and the sum of the integral of percentage shared in the vector line segments, capable of reducing the accumulated tolerance effect of each line segment node, to acquire better approaching quality. Together with the point-to-point approaching process means from the tailing node to the heading node of the set of line segments, the minimum vector line segments of vector font can be acquired to achieve the goal of data amount reduction.

Description

46 48 3 0 V. Description of the invention (1) [Scope of application of the invention] The present invention has n ^ off ~ species for generating digital orientation I_M_7, especially a kind of target for sweeping γ and M_M2. The method of transforming to the child of I, near processing, and then produces a set of polyline segments obtained by ί to perform a close processing method [Background of the invention]. At present, for the production method of vector fonts, or the font mode bitmap is used as the rotation, the absence of A Danzi body handle 1 meaning and two inputs, the description of the vectorized processing body 'is the vector line that constitutes this vector font. Bailizi's processing process includes the following steps: Scan the font model (or font type bitmap) to obtain—a set of polyline segments that form the outline of the font. This set of polyline segments includes many polyline segments that end to end, and each one The polyline segment is made up of many pixels next to each other, and each polyline segment is defined by the line connecting the last pixel of the first pixel spot (hereinafter, we will refer to the α polyline node Call the first pixel and the last pixel of the polyline segment); Set the polyline ^ and set 'approximation processing with a straight line or a 3rd B-degree curve to obtain the descriptive data of the font (point coordinates)' The processed font description data includes the following three types of descriptions:. Straight line description data: contains U〇, y0) (XI, yl) 2 point coordinates. Third degree B-curve description data: contains (X 0, y 〇). ), (X 1, y 丨), (-X 2 ′ y 2), (X 3 ′ y 3) 4 point coordinates. Rectangular description data: (x〇, y〇) (XI, yl) 2 points—coordinates.

Page 4 46 48 3 0

V. Description of the invention (2) For the second step of making a vector font, the set of polyline segments obtained after the type is described by approximation of the hand form (straight green, third degree B-curve, rectangle)- ^ ^ Yilianghe. So 'approach processing is accurate, # directly most dependent ^ Qian's aesthetics and vector when displayed as a vector font === Take the polyline segment set ABC shown in Figure 1' 'as an example (in the next handover) According to the judgment of the naked eye, 'the best approximation descriptive data should include: descriptive data of the straight line' 3rd degree B-curve & descriptive data. Please read the "Figure 2" one by one, which is the line segment of the conventional technique. Set the principle diagram of cooperative line approximation. A set of polyline segments obtained after scanning the font dot matrix is set as' first node is A and tail node is ε. Please refer to "Figure 3" for approximate approximation of straight lines. The steps are roughly As follows: 1. Take the first node A of the set of polyline segments as the starting point, and try to find the points one by one in the increasing direction towards the tail node e. As shown in "Figure 3", the testable point is B ', then 丨 A' B 丨Is a current set of nodes. 2. The current set of tentative nodes is set to the current vector to approximate the straight line. 3. Calculate the average approximation error, which is the straight line approximation error: Endpoint approximation error: that is, (A, B 丨A node in a point set Vector approximation straight vertical line, the vertical line is the length of this section polylines - point vector approximation straight line approximation error.

464830 V. Description of the invention (3) Average approximation error: Set the approximation error value of each node to the vector approximation line as Φ i, then the average approximation error is the mathematical average of the approximation errors of all nodes in the set of points A and B Value, as shown in formula (1): Φ = (ΦΑ + ΦΒ) / 2 · · * Formula (1) When dKOLine ((DLine is the allowable error value of the preset straight line), continue to test forward the next Node. Taking the "Figure 2" as an example, you can continue to test {A, B, C} ....... {A, B, C, D} and other node sets and repeat step 1. When Φ > Φ At the time of L ine, the test stops, and the first line to the last node in the node set at the time of Φ < Φ L ine but closest to Φ L 1 ne at that time is used as the description of the straight line approximation. Take "Figure 2" as For example, when U, B, C, D} is approached, the approximation error exceeds Φ L ne, then a straight line segment description [AB-C1 (herein, the symbol "1" is used to indicate the approximated vector line segment) is obtained, with a total of 3 nodes. However, when the approximation processing of the broken line segment is actually performed, the operation of straight line approximation and curve approximation will be performed at the same time. This is the last approximation description. The approximate processing steps are shown in Figure 4; (step a5) is to determine which of the straight line description data and the curve description data have the same starting point of the vector. Contains the most polyline nodes, and decides to use a relatively large number of polyline nodes as the vector line segment description information. This method will minimize the number of vector line segments constituting the vector font to achieve the purpose of saving the data capacity of the vector font. Obviously, the conventional method has two disadvantages:-1. Approximation error of the polyline segment: it is the number of approximation errors of all nodes

V. Description of the invention (4) The average value of learning ’It obviously ignores that the influence of each node in the entire approximation process is different. For a set of polyline segments to be approximated, a total of n nodes are set: P1 ′ P2 ′ P3 ′ P4,… Pη, assuming that the current line / curve approximation is used to calculate the approximation error from P1 to P η. At this time, the entire node set is {PI, P2, P3, P4 ... Pη} ', then the entire approach avoidance error Φ = (+ ΦΡ2 + ΦP3 + ΦP4 +… ΦΡη) / η; take "Figure 5" as an example to illustrate this obvious Problem: Let the entire polyline & set be [A, β, c], then the approximation error Φ 'from node a to node c will be calculated as formula (2) according to the conventional principle: Φ = (ΦΑ + φΒ + φ〇 / 3 *. · (Equation 2); where the error must be 0 for the head and tail nodes of A 'C, that is, φ Α = 0, φ〇〇, so φ = φΒ / 3. In this approach, it is possible There is a result of φ > φ Line, and the approximation is described by 2 straight lines [0], but in fact, the error of this approximation is a poor result when viewed by human eyes. If you notice that the line segment is only in the The entire vector line segment ^ accounts for a small proportion, and when it has little effect on the entire approximation, it should be regarded as approximation into one line segment. The obvious result is obvious. If we simply simply calculate the approximation error of each node, The average value is used as the basis for the determination, and it still does not accurately reflect the local information in the set of polyline segments to the entire approximation The impact of vector line segments. 2 'heuristic approximation conventional way by an incremental point tempted direction' of the noise (η〇 Shu c e) greatly affects the error approximations. H and the word quality nearly affected.

Page 7 4 6 4 8 3 0_. V. Description of the Invention (5) Here we will take “Figure 6” as an example to explain, suppose the set of broken line segments A-B-C-D-E-F is used for approximation processing. When the traditional incremental approach heuristic method is used, the heuristic directions are sequentially A, B, C, D, E, F and other nodes. Take "Figure 6" as an example. At the B and C nodes, and at the D and E nodes, there is a noise error of the font font. Obviously, a straight line approximation is performed. When the preset allowable value Φ L ine of the straight line approximation error is small, The noise error at the node set {B, C}, {D, E} and the cumulative error to a certain extent may cause Φ > Φ L 1 ne, so that the approximation result of the set of polyline segments will be described by several approximation straight lines . But in fact, with human eyes, the better approximation result should be A-F as a whole approximation line. Obviously, the purpose of the approximation process is to compare the description line with the original set of polyline segments. On a macro scale, try to ensure similar approximation. However, when approaching the first node of the set of polyline segments in increasing direction, it is likely to be in a certain area because of noise errors. "Only local, but not the beginning and end M. Many of the better results that should have been an approximation line, wrong approximation. There are several approximation lines. Therefore, the traditional fitting method is to start from the first endpoint of the polyline segment, and gradually increase the nodes of the polyline segment set under the premise of meeting the preset allowable error value, until it exceeds the error range or the entire polyline The segment set ends. In the process of curve fitting, the deviation of individual points at the beginning will cause the operation of the approximation to end, resulting in a curve being approximated by several straight or curved segments, increasing the number of nodes and affecting the approximation effect. At the same time, the content of the information has been increased. [The purpose and summary of the invention]

Page 8 8 3! V. Description of the invention (6) The main purpose of the present invention is to improve the inaccuracy of the conventional approximation processing method and the verbose lack of data. For this reason, the present invention proposes to multiply the approximation error at each node by the product of the corresponding straight line segment in the entire vector line segment, and then add the total value as the approximation error of the judgment, which can be refined. It reflects the approximation error of each detail of the entire set of polyline segments. On the other hand, the present invention performs a point-by-point approach to decrease the tail node of the set of polyline segments toward the head node in a decreasing direction, and will obtain the minimum number of vector line segments that constitute a vector font, avoiding the redundant phenomenon and reducing the amount of data. The detailed technical content of the present invention and its embodiments are described below with reference to the drawings. [Brief description of the drawings] Figure 1 * is a legend of the set of polyline segments obtained by the conventional scanning font model. ./Figure 2 is a legend illustrating the principle of linear approximation. _Figure 3 is a flow chart of the incremental approximation of the maximum approximation line / curve used to obtain the set of polyline segments. Fig. 4 is a basic flowchart of the approximation processing for a set of polyline segments. Fig. 5 is an illustration of a set of polyline segments that illustrate the error of the approximation error in the conventional approximation process. v Figure 6 is an illustration of a set of polyline segments that will cause noise error when the point-by-point approximation in the increasing direction is used in the conventional approximation process.

Page 9 46 48 3 〇 V. Description of the invention (7) Figure 7 is a diagram explaining the principle of calculating the approximation error according to the present invention. Fig. 8 is a flowchart of incremental approximation of a straight line / curve for obtaining the maximum approximation of a set of polyline segments according to the present invention. Fig. 9 is a partial flowchart of the present invention. Fig. 10 is a diagram illustrating an embodiment of processing a set of polyline segments by point-by-point approach in a decreasing direction using the method of the present invention. [Detailed description of the preferred embodiment] First, please refer to "Figure 7". Assume that a set of polyline segments to be approximated includes η polyline nodes: PI, P2, P3, P4, .... * Pη 'False 6 and At present, a straight line / curve approximation is performed, and the approximation error of each polyline node from the first node P1 of the polyline to the tail node ρη of the polyline is calculated. P4 .. · P: When the node set of the entire polyline segment set is ίPI, P2, P3, 1. First, make the first node of the polyline P1 to 挢 ^ This connection Hi] is the current approximation of the vector ^ tail 卽 point Pn Line, let it be the vector of the vector line segment [Pi-pn]: 'Where the first node of the polyline? 1M ... Pnq point to the Dun section ρι_Ρη and the polyline node P2, and the line intersection point is P2', P3 ,, ... ..X straight line, the relative corresponding close error value is the approximation of each polyline node from the polyline node to its Π relative. ... the length of the line where the perpendicular intersects

Then from the set of P1, ρ2, P3 · _'Pη-i, P, the polyline nodes of each polyline node in the entire polyline segment of Xiangdan can be used in the vector line segment [P ΡΡη wide, f segment [] t -The shadow ... 4 6 4 3 3 0

5. Description of the invention (8). The following uses W (Px) n ~ n to denote that each W (P1) = the length of the line segment / the length of the line segment [Pl-Pn]) 20; ^ 'long night (that is to Θ W (P2) = line segment £ 1 : Length of ^ / segment complex W (P3) = length of segment / length of segment; ......... ~~% length; W (Pn-1) = length of segment / segment ρι_ρη W (Pn) = Length of line segment / line length ^, length;; ~~ negative degree = j 2. Then the approximation error point W (Px) of the entire approximation vector line segment [Pi ^ -Pn] is multiplied by the approximation error of the corresponding polyline node. 3. The difference can be accurately described by the pair. Its calculation formula is shown in formula (3). The approximation error of the total approximation of the vector line segment [E1z £ q]: φ—φ W (P1) + ㈣2 " ( Ρ2) + ΦΡ3 * W (P3) +… φρη_ " W (Pn-l) + φρη * w (Pn).. _ ·. · Formula (3); Obviously, this expression is better than the conventional method by The mathematical average description of the approximation error is much more accurate. The purpose is to accurately describe the effect of the approximation error of each polyline node on the polyline segment on the entire approximated vector line segment [£ lz £ n]. P, Γ > right! ^ Take "Figure 2" as an example: The set of segments is [A, B. The approximation error F of the polyline node A to the polyline node C is now calculated. It is processed according to the method of the present invention, and its formula is _ Φ ΦΑ * (ΑΑ / AD) + ΦΒ ^ (αβ '/ AD) + OC ^ (AC! /

4 6 4b ^ 〇 --------- __ V. Description of the invention (9) AD) + 〇D * (AD / AD); Among them, for A and D nodes, the error must be 0, Φ A = 0, ΦΓ) = 〇. Please refer to "Figure 8" again is a flowchart of the steps for obtaining the maximum approximation line / curve of the set of polyline segments in an increasing direction according to the present invention, which includes: Initializing each parameter value, including: the limit value OLine of the allowable error, set The first node in the set of polyline segments = Start, the tail node in the set of polyline segments = End, and the vector approximation point (Current Point) of the current vector line segment is End (where the first node is the vector of the current vector line segment (Starting point); determine whether the Current Point is Start? If yes, go to step 8, otherwise go to the next step; according to the method of the present invention, 'calculate the sum of the approximation error φ of the straight line / curve from Start to Current Point; judge Is Φ greater than Φ L ine? If yes, go to the next step, otherwise go to step 6; Set Current Point = Current Point-1 (that is, from C urre η 1; P oint decreases in the direction of Start A polyline segment node), and return to step 2; let the vector approximation point of the better approximated vector segment be Current Point (that is, the end point of the vector segment is Current Point);

Current Point = Start, and returns to the previous step 2 (that is, the polyline node corresponding to the better vector approximation point is updated to the vector

Page 12 4 6 κ c V. Description of the invention (10) Start point, repeat the above step 2 to determine the approximation vector segments of the remaining polyline nodes until all polyline nodes are processed); Set the approximated vector line segments The vector approximation point is start; the approximation process ends. Of course, in the above-mentioned processing steps of the present invention, it is still necessary to perform the approximation processing on the set of polyline segments as shown in "Figure 4", and simultaneously perform straight-line and curve approximation operations, and select the better one from among them as the final solution. The approximation description is sought, so the better vector approximation point referred to in this step 6 is: from the starting point of the vector to the better vector approximation point, the one with a larger number of polyline nodes is covered. Therefore, the above step 6 also includes at least the following steps (as shown in "Figure 9"): 6. 1 Calculate the vector approximation from the vector starting point to the straight-line approximation of the total error value 17 which is less than and closest to the limit of the allowable error The number of polyline nodes (L ine N um) covered between the points; 6. 2 Calculated between the vector approximation points from the vector starting point to the total error value of the curve approximation A is less than and closest to the limit of the allowable error The number of polyline nodes (Curve N um); and 6.3 The vector approximation point that covers the larger number of polyline nodes is the Current Point (that is, the better vector approximation point). Finally, please refer to "Fig. 10", which uses a set of broken line segments ABCDEF as an example to describe the processing process of the method according to the present invention; the technique of performing point-by-point approximation processing according to the present invention in a decreasing direction, to the node that first performs the approximation process The set system is [A, B, C, D, E, F].

Page 13 464830

Page 14

Claims (1)

d 6 4 b-) t. Patent application scope 1. A digitalized vector font approximation processing method for processing a set of polyline segments containing a plurality of polyline nodes into a digitalized vector font, including: determining an allowable error Limit value; the first node of the set of polyline segments is the vector starting point of the currently approaching vector line segment; the direction of the tail node of the set of polyline segment sets is decreasing toward the first node, and the polyline node is set point by point as the vector of the current vector line segment. Point; from the vector approaching point in the decreasing direction to the starting point of the vector, the product of the approximation error of each polyline node that is included in the current vector line segment and the proportion value of the proportion value in the current vector line segment is calculated point by point. Total error value; Find the vector approximation point where the total error value is less than and closest to the limit of the allowable error, and determine a better vector approximation point from the vector approximation point; use the better vector approximation point The corresponding polyline node is updated to the starting point of the vector, repeating the previous point-by-point set in the decreasing direction as the vector approximation point of the current vector line segment Step, until the vector approximation point for the first node until the node of the polyline. 2. The approximation processing method of the digitized vector font as described in item 1 of the scope of the patent application, wherein the approximation error of each vector node in the current vector line segment includes the straight line approximation error from the starting point of the vector to the vertex of the polyline. Approximation error with curve. — 3. Approximation of digitized vector font as described in item 2 of the scope of patent application Page 15 4 6 4 8 0 6. The method of patent application range, where the straight line approximation error refers to the shortest vertical distance from the line node to a line connecting the starting point of the vector to the trailing node of the line segment. 4. The approximation processing method for the digitized vector font as described in item 1 of the scope of patent application, wherein the proportion value of each vector node in the current vector line segment in the current vector line segment is obtained by the following steps : Obtaining the length of the line from the starting point of the vector to the trailing node of the polyline segment; obtaining the intersection of the vertical line from the line node to the connecting line at the connecting line; obtaining the length of the straight line from the starting point of the vector to the crossing point; The calculation formula (straight line length / line length) is calculated to obtain the proportional value. 5. The digitized vector font approximation processing method described in item 1 of the scope of patent application, wherein the total error value includes: the total error value of the linear approximation and the total error value of the curve approximation. 6. The digitized vector font approximation processing method as described in item 1 of the scope of patent application, wherein the preferred vector approximation point is any of a linear approximation point or a curve approximation point. The approximation processing method of the digitized vector font described in the first item of the patent scope, wherein the better vector approximation point is: from the starting point of the vector to the better vector approximation point, the number of polyline nodes is covered. . Page 16