JPH0243220B2 - - Google Patents
Info
- Publication number
- JPH0243220B2 JPH0243220B2 JP55189130A JP18913080A JPH0243220B2 JP H0243220 B2 JPH0243220 B2 JP H0243220B2 JP 55189130 A JP55189130 A JP 55189130A JP 18913080 A JP18913080 A JP 18913080A JP H0243220 B2 JPH0243220 B2 JP H0243220B2
- Authority
- JP
- Japan
- Prior art keywords
- distance
- gravity
- center
- character pattern
- character
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/18—Extraction of features or characteristics of the image
- G06V30/182—Extraction of features or characteristics of the image by coding the contour of the pattern
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Character Discrimination (AREA)
Description
【発明の詳細な説明】
本発明は、文字認識に係り、文字の特徴抽出方
式に関する。DETAILED DESCRIPTION OF THE INVENTION The present invention relates to character recognition, and more particularly to a character feature extraction method.
文字認識は、各種の文字の特徴の抽出方式によ
つて得られた情報にもとづいて、当該特徴を有す
る文字を特定することによつて行なわれる。この
際、文字の特徴の抽出方式としては各種の手段が
従来より採用され文字を正確に認識するために
は、前記手段を多種類必要とする。 Character recognition is performed by identifying characters having the characteristics based on information obtained by various character feature extraction methods. At this time, various means have been conventionally employed as a method for extracting the characteristics of characters, and in order to accurately recognize characters, many types of means are required.
本発明は、上述の点に鑑みて、文字の特徴を抽
出する新規な特徴抽出方式を提供することを特徴
とする。 In view of the above points, the present invention is characterized by providing a novel feature extraction method for extracting character features.
本発明の特徴抽出方式は文字パターンを入力す
る手段と、該入力した文字パターンの領域に対し
て複数の方向において走査線をそれぞれ順次文字
パターンの方に文字パターンを検出するまで移動
し、該文字パターンを検出した時の走査線を辺と
する前記文字パターンの外接多角形を求める外接
多角形生成手段と、前記入力した文字パターンの
重心位置を求める重心位置検出手段と、該重心位
置検出手段で求めた重心位置を中心とした距離画
を作成する距離発生回路と、前記文字パターンの
重心点より複数方向に放射した線が前記外接多角
形と交わる点の位置を求めるとともに、前記距離
画より前記重心点からの距離を計測する計測手段
とからなり、前記距離から文字パターンの特徴を
検出することを特徴とする特徴抽出方式である。 The feature extraction method of the present invention includes means for inputting a character pattern, and scanning lines are sequentially moved toward the character pattern in a plurality of directions with respect to the area of the input character pattern until the character pattern is detected. a circumscribed polygon generating means for calculating a circumscribed polygon of the character pattern whose sides are the scanning lines when the pattern is detected; a barycenter position detecting means for determining the barycenter position of the input character pattern; A distance generation circuit that creates a distance drawing centered on the obtained center of gravity position, and a distance generation circuit that calculates the position of a point where a line radiating from the center of gravity of the character pattern in multiple directions intersects the circumscribed polygon, and calculates the distance drawing from the distance drawing. This feature extraction method is characterized by comprising a measuring means for measuring the distance from the center of gravity, and detecting the features of the character pattern from the distance.
以下、図面を参照して、本発明の一実施例を説
明する。 Hereinafter, one embodiment of the present invention will be described with reference to the drawings.
第1図に示された本発明の特徴抽出方式は、文
字入力が加えられる外接多角形生成回路1と、文
字入力が加えられる重心位置検出回路2−1と距
離発生回路2−2とよりなる距離画作成回路2
と、外接多角形生成回路1と距離画作成回路2の
出力が入力される特徴生成回路3とを基本的構成
として含むものである。 The feature extraction method of the present invention shown in FIG. 1 includes a circumscribed polygon generation circuit 1 to which character input is applied, a barycenter position detection circuit 2-1 to which character input is applied, and a distance generation circuit 2-2. Distance image creation circuit 2
The basic configuration includes a feature generation circuit 3 to which the outputs of a circumscribed polygon generation circuit 1 and a distance image generation circuit 2 are input.
文字入力端子T1に加えられる、たとえば、第
2図aに示される如き入力文字「永」は、布線論
理もしくはマイクロプロセツサで構成された外接
多角形生成回路1において、第2図bに示される
ように、その外接多角形10を発生するように処
理される。すなわち、アドレス空間を矩形領域1
1として扱い該領域11に入力文字「永」を第2
図aに示すようなデータパターンとして格納し、
アドレス空間矩形領域11の辺に沿つて複数の点
101〜10kを走査の出発点としあらかじめ指定
してある方向の走査を点101から10kに走査線
が順次矩形領域11の内部に移動するように選定
し走査する。各出発点から走査する線は互いに平
行な線に沿つて走査していつて入力文字「永」を
最初に検知した走査線11kを外接多角形10の
1辺とする。以下同様の処理をあらかじめ定めら
れた数だけ行うことにより外接多角形の辺の数だ
け繰り返することとなる。前述の動作処理によつ
て求めた「永」を検出した各線が外接多角形10
を生成する。尚、入力文字「永」に対する走査線
のアクセスは、走査線がx方向あるいはy方向の
場合は、カウンタの出力をアドレスとして行えば
よい。第2図aに例示したように斜め方向の走査
線によるアクセスの場合は、ある走査線に注目す
るとアクセスするアドレスの変化は規則的なの
で、開始アドレス、アドレス変位方向、終了アド
レスをあらかじめメモリに記憶しておけば、比較
的容易にアクセスするアドレスを発生できる。 For example, the input character "ei" as shown in FIG. 2a, which is applied to the character input terminal T1 , is input as shown in FIG. It is processed to generate its circumscribed polygon 10 as shown. In other words, the address space is rectangular area 1
It is treated as 1 and the input character “ei” is inputted into the area 11 as the second character.
Store it as a data pattern as shown in figure a,
Scanning starts from a plurality of points 10 1 to 10 k along the sides of the address space rectangular area 11 and scans in a pre-specified direction from point 10 1 to 10 k sequentially inside the rectangular area 11. Select and scan to move. The lines scanned from each starting point are parallel to each other, and the scanning line 11k where the input character "ei" is first detected is defined as one side of the circumscribed polygon 10. Thereafter, similar processing is performed a predetermined number of times, thereby repeating the same number of times as the number of sides of the circumscribed polygon. Each line that has detected “eong” obtained through the above-mentioned motion processing is a circumscribed polygon 10
generate. Incidentally, when the scanning line is in the x direction or the y direction, the input character "ei" can be accessed by using the output of the counter as an address. In the case of access using diagonal scanning lines as illustrated in Fig. 2a, since the address to be accessed changes regularly when focusing on a certain scanning line, the start address, address displacement direction, and end address are stored in memory in advance. If you do this, you can generate an access address relatively easily.
次に、アドレス空間の矩形領域11の任意の1
点としてたとえば、入力文字「永」の重心0を求
め、この重心0より、第3図aに示すように、外
接多角形10の辺上の複数の点までの距離d1…
d12を計測する。文字「永」の重心0のx座標は、
画素のx座標の和をとり、画素数で割ることによ
り求められる。同様にして重心0のy座標も求め
ることができる。 Next, any one in the rectangular area 11 of the address space
For example, find the center of gravity 0 of the input character "ei" as a point, and calculate the distances d 1 from this center of gravity 0 to multiple points on the sides of the circumscribed polygon 10, as shown in Figure 3a.
Measure d 12 . The x-coordinate of the center of gravity 0 of the character “ei” is
It is obtained by taking the sum of the x-coordinates of pixels and dividing by the number of pixels. Similarly, the y-coordinate of the center of gravity 0 can also be determined.
距離画作成回路2内の重心位置検出回路2−1
はこの重心位置を求める回路であり、入力文字
「永(文字画像)を一時記憶し、ドツトの存在す
るX座標及びY座標を順次加算するとともにドツ
ト数を計算し、全ドツト数終了時にX座標及びY
座標の加算結果を係数結果で割り、その結果を重
心位置とする。 Center of gravity position detection circuit 2-1 in distance image creation circuit 2
is a circuit that calculates the position of the center of gravity, which temporarily stores the input character ``English'' (character image), sequentially adds the X and Y coordinates where dots exist, calculates the number of dots, and calculates the X coordinate when all dots are counted. and Y
Divide the coordinate addition result by the coefficient result and use the result as the center of gravity position.
距離Aの計測は第1図に示した距離画作成回路
2の距離発生回路2−2によつて第3図bに示し
た3×3ウインド処理回路20を用いて行う。こ
のウインド処理回路20は、第3図bに示すよう
に、アドレス空間矩形領域11のx軸方向のセル
(単位領域)数と等しい数のセルを有するシフト
レジスタ21を3段直列接続し、アドレス空間矩
形領域11からの画素データをシフトレジスタ2
1に入力する。3段のシフトレジスタのうち、3
×3個のセルを用いてウインド22を形成する。 The distance A is measured by the distance generation circuit 2-2 of the distance image creation circuit 2 shown in FIG. 1 using the 3.times.3 window processing circuit 20 shown in FIG. 3b. As shown in FIG. 3b, this window processing circuit 20 has three stages of shift registers 21 connected in series, each having a number of cells equal to the number of cells (unit area) in the x-axis direction of the address space rectangular area 11. Pixel data from the spatial rectangular area 11 is transferred to the shift register 2.
Enter 1. 3 of the 3 stage shift registers
The window 22 is formed using x3 cells.
予め重心点0はフラツグビツトを1、他を点の
フラツグビツトは0としておく。シフトレジスタ
にはこのフラツグビツトを入力する。 In advance, the flag bit of the center of gravity 0 is set to 1, and the flag bits of the other points are set to 0. This flag bit is input to the shift register.
先ずウインドを走査(全画面に渉つて移動)し
中心Aのセルが0で周囲8点のうち少くとも1点
が1であればAに相当する画素を1と定義し同時
にフラツグビツトを1にする。これにより第3図
cの8個の1が定義される。 First, scan the window (move across the entire screen), and if the cell at the center A is 0 and at least one of the surrounding 8 points is 1, define the pixel corresponding to A as 1, and at the same time set the flag bit to 1. . This defines the eight 1's in FIG. 3c.
次にウインドをもう一度走査し、同様の論理で
2の点を定義し、フラツグビツトを1にする。 Next, scan the window again, define point 2 using similar logic, and set the flag bit to 1.
以下同様にてウインド走査を合計n回繰り返す
ことにより第3図cのように1〜nと定義された
画素より成る距離画が得られる。次に特徴生成回
路3によつて前記距離画において外接的多角形1
0の辺に対応する画素に定義された数たとえばl
より、重心0から外接多角形10までの距離dlを
得る。重心0より外接多角形10の複数の点まで
の距離dを複数個計測することにより入力文字
「永」の特徴を第3図dに示すように距離の列と
して得る、そして、この距離の列を用いて、入力
文字「永」の特徴を抽出する。 Thereafter, by repeating the window scan a total of n times in the same manner, a distance image consisting of pixels defined as 1 to n as shown in FIG. 3c is obtained. Next, the feature generation circuit 3 generates a circumscribed polygon 1 in the distance picture.
The number defined for the pixel corresponding to the 0 side, for example l
From this, the distance dl from the center of gravity 0 to the circumscribed polygon 10 is obtained. By measuring a plurality of distances d from the center of gravity 0 to a plurality of points of the circumscribed polygon 10, the characteristics of the input character "ei" are obtained as a string of distances as shown in FIG. Extract the features of the input character ``ei'' using .
上述したように、本発明によれば、入力文字に
対してアドレス空間の矩形領域において外接多角
形を生成し、任意の1点より外接多角形の辺上の
複数の点までの複数の距離情報を計測して文字の
特徴を抽出するもので、この特徴抽出方式によつ
て、文字認識をより正確に行えるようにできるも
のである。 As described above, according to the present invention, a circumscribed polygon is generated in a rectangular area of an address space for an input character, and a plurality of distance information from an arbitrary point to a plurality of points on a side of the circumscribed polygon is generated. This method measures character characteristics and extracts character features.This feature extraction method allows for more accurate character recognition.
第1図は本発明の特徴抽出方式の基本的構成を
示すブロツク図、第2図aはアドレス空間矩形領
域に形成された入力文字と外接多角形生成方法を
示す画像図、第2図bは同じく入力文字と外接多
角形の画像図、第3図aはアドレス空間矩形領域
における外接多角形の辺と任意の点との距離を示
す図、第3図bは3×3ウインド処理回路のブロ
ツク図、第3図cは前記ウインド処理回路によつ
て形成される距離画、第3図dは第3図aに示し
た複数の距離よりなる距離別である。
1……外接多角形生成回路、2……距離画作成
回路、3……特徴生成回路、10……外接多角
形、111〜11k……走査線、11……アドレス
空間矩形領域、21……シフトレジスタ、22…
…3×3ウインド処理回路。
FIG. 1 is a block diagram showing the basic configuration of the feature extraction method of the present invention, FIG. Similarly, an image diagram of an input character and a circumscribed polygon is shown. Figure 3a is a diagram showing the distance between a side of a circumscribed polygon and an arbitrary point in a rectangular address space area. Figure 3b is a block diagram of a 3x3 window processing circuit. 3c is a distance image formed by the window processing circuit, and FIG. 3d is a distance image formed by a plurality of distances shown in FIG. 3a. 1... Circumscribed polygon generation circuit, 2... Distance image generation circuit, 3... Feature generation circuit, 10... Circumscribed polygon, 11 1 to 11 k ... Scanning line, 11... Address space rectangular area, 21 ...Shift register, 22...
...3x3 window processing circuit.
Claims (1)
方向において走査線をそれぞれ順次文字パターン
の方に文字パターンを検出するまで移動し、該文
字パターンを検出した時の走査線を辺とする前記
文字パターンの外接多角形を求める外接多角形生
成手段と、 前記入力した文字パターンの重心位置を求める
重心位置検出手段と、 該重心位置検出手段で求めた重心位置を中心と
した距離画を作成する距離発生回路と、 前記文字パターンの重心点より複数方向に放射
した線が前記外接多角形と交わる点の位置を求め
るとともに、前記距離画より前記重心点からの距
離を計測する計測手段とからなり、 前記距離から文字パターンの特徴を検出するこ
とを特徴とする特徴抽出方式。[Scope of Claims] 1. means for inputting a character pattern; a means for inputting a character pattern; a circumscribed polygon generating means for calculating a circumscribed polygon of the character pattern whose sides are scanning lines when the pattern is detected; a center of gravity position detection means for determining the position of the center of gravity of the input character pattern; a distance generating circuit that creates a distance drawing centered on the obtained center of gravity; a distance generation circuit that calculates the position of a point where a line radiating in multiple directions from the center of gravity of the character pattern intersects the circumscribed polygon; A feature extraction method comprising a measuring means for measuring a distance from a center of gravity, and detecting features of a character pattern from the distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP55189130A JPS57113183A (en) | 1980-12-29 | 1980-12-29 | Feature extraction system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP55189130A JPS57113183A (en) | 1980-12-29 | 1980-12-29 | Feature extraction system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| JPS57113183A JPS57113183A (en) | 1982-07-14 |
| JPH0243220B2 true JPH0243220B2 (en) | 1990-09-27 |
Family
ID=16235896
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP55189130A Granted JPS57113183A (en) | 1980-12-29 | 1980-12-29 | Feature extraction system |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPS57113183A (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH03223986A (en) * | 1989-12-26 | 1991-10-02 | Fuji Facom Corp | Method for rejecting recognized result |
| US5054094A (en) * | 1990-05-07 | 1991-10-01 | Eastman Kodak Company | Rotationally impervious feature extraction for optical character recognition |
-
1980
- 1980-12-29 JP JP55189130A patent/JPS57113183A/en active Granted
Also Published As
| Publication number | Publication date |
|---|---|
| JPS57113183A (en) | 1982-07-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dyer et al. | Parallel image processing by memory-augmented cellular automata | |
| JP3400151B2 (en) | Character string region extraction apparatus and method | |
| JPH0823890B2 (en) | Method and apparatus for automatically verifying signature | |
| US8005262B2 (en) | System and method for video object identification | |
| JP3471578B2 (en) | Line direction determining device, image tilt detecting device, and image tilt correcting device | |
| JPH0243220B2 (en) | ||
| US6373997B1 (en) | Coarse and fine skew measurement | |
| EP0702320A1 (en) | Skew detection | |
| JPH0750496B2 (en) | Image signal processor | |
| JP2882314B2 (en) | Line figure recognition method | |
| JPS646508B2 (en) | ||
| JPH01205287A (en) | Character line inclination detector | |
| KR100295245B1 (en) | A parallel thinning method using weighted value | |
| JPH0799532B2 (en) | Character cutting device | |
| JP2616994B2 (en) | Feature extraction device | |
| JP2745911B2 (en) | Image processing device | |
| JPS5914782B2 (en) | Binary pattern digitization processing method | |
| JP2827288B2 (en) | Character recognition device | |
| JPS6252911B2 (en) | ||
| JP2002109544A (en) | Seal impression positioning method and device | |
| CN114972044A (en) | Point cloud denoising method, device, electronic device and storage medium | |
| EP1439485A1 (en) | Segmenting a composite image via basic rectangles | |
| JPS63238686A (en) | Feature extracting system | |
| JP2002251613A (en) | Graphic recognition method and apparatus | |
| JPS6343788B2 (en) |