JP4884612B2 - Surveyor equipped with an automatic character reading device - Google Patents

Description

【００６２】
パターン１０２では無く数値で表示している標尺の場合は、Ｓ２７の小さな文字の認識と同様に処理して、数値を認識する。
【００６３】
パターン１０２については、その存在範囲を決定するときに、大きな文字と目盛りの位置からその存在範囲を決定しても、多少のズレが発生する可能性がある。パターンを認識する方法として、文字の認識と同様に、その存在範囲内の一定比率で決定される特徴点のレベルで判断する方法が考えられるが、その存在範囲のずれによって間違った認識をする可能性がある。この誤認識を防止するためにパターンの存在範囲を修正することを考慮すると、パターンの存在範囲内のデータレベルを確認し、データレベルが変化する点を検出して、それらの最も外側に位置する部分を存在範囲の端と判断することで、パターンの存在範囲を修正することが可能である。しかし、このような方法でパターンの存在範囲を修正した場合、パターンの外形の寸法はそれぞれ異なるので、特徴点はさらにずれて、間違った認識をする可能性が増加する。
そこで本実施形態では、データの中から、まず確実にパターンの存在を確認できる列のレベル変化の仕方を確認し、判別することで、認識率を上げることを可能にした。
【００６４】
Ｓ２５： 目盛りの数を求める（１０mmの位を求める）
（5-1） 大きな数字の上辺ＵＣと求める位置（水平ライン５３、アッパスタジアライン５４またはロアスタジアライン５５）との間に含まれる画素数を算出する。
（5-2） 求めた画素数を１０mm目盛りの１周期当たりに含まれる画素数で割り、その整数部分だけを取り出して、目盛りの個数を求める。
【００６５】
大きな数字の位置から測定する位置までの目盛りの数を求めるには、目盛りの数を計数すればよいが、これを単純に処理するために、大きな文字の上辺の垂直座標を初期位置として同一垂直方向列上の画素データを順に確認しながら白から黒、黒から白に変化する変化点を上記測定する位置まで検出し、上記測定範囲中に一目盛分の変化点が何個存在するかを計数すればよい。
【００６６】
ただし、この方法では測りたい間隔にあるすべての画素データにおいて、直前の画素データと今回の画素データのレベルを比較して変化点を検知する必要がある。
これに対して文字の上端の位置と測定したい位置のエリアセンサ上の位置情報（座標）をお互いの位置の座標の差をとると、２つの位置の間に含まれる画素数が求められる。その画素数を１周期当たりに含まれる画素数で割り、その商の整数部分だけを取り出せば、目盛りの数が求められる。文字の上端と求める位置が極端に近い場合を除けば、この処理の方が短時間で済む。
【００６７】
Ｓ２６： 任意の辺の長さが所定のしきい値よりも大きい場合（小さな文字列１０４の文字の位置を検出する）
（6-1） 標尺の右端位置、および目盛りの水平、垂直方向の長さを基準として、小さい文字が存在する範囲を求める。
目盛りの水平方向の長さに対する目盛りの右端位置からの各小さい文字までの長さを前もって記憶しておき、目盛りの水平方向の辺に含まれる画素数を数え、それに比率をかけて各文字の水平方向の位置（文字の右端と左端）を求める。
鉛直方向は、目盛りの黒レベルの上端と文字の上端が一致し、下端は、目盛りの一周期に相当する画素数に、比をかけることで求めることができる。
（6-2） （6-1）で求めた端の位置について、それぞれ四方の辺にあたるデータを確認して黒レベルが連続して存在する列としない列が隣り合わせている二列を検出して、このときの黒レベルが連続して存在する列を文字の端とする。
文字の端とは、小さい文字が内接するように囲んだ四角形のそれぞれの辺の位置のことであり、小さい文字の検出はこれらの四辺の位置をすべて求めることである。
【００６８】
このように小さな文字の検出処理では、大きな文字の位置検出と同様にして、始めに大まかな位置を求め、そこから文字の範囲を絞り込むことで、検出時間の短縮と正確な切り出しを両立させている。
【００６９】
Ｓ２７ ３桁の小さな文字の認識（1,000mm、100mm、10mmの位の数字をそれぞれ認識する）
（7-1） Ｓ２６で求めた小さな文字の位置情報から小さな文字の縦横の比を求め、その結果から文字が「１」かそれ以外の数字かを認識する。この縦と横の比率は、「１」の場合は約５：１、それ以外の場合は約３：１になるので、４：１を基準として縦横比がそれより大きければ１と判断し、それ以下であれば「１」以外であると判断する。
（7-2） （7-1）において「１」と認識されなかった場合、予め記憶しておいた、図８のように数値が確実に認識できる位置の画素データを、予め設定された順番に認識する。つまり、各位置の画素データのレベルを確認する毎に、その情報の組み合わせで決まる数字を決定していく方法で文字を認識する。
【００７０】
本実施の形態では、縦横比が４：１以下の場合に、小さい文字の左上端を基準位置として、以下の順に認識する。図９および図１０は、「１」以外の数字を認識するアルゴリズムを示す模式図およびフローチャートである。図９において、縦長の長方形が小さな文字の切り出し範囲（輪郭）であり、％で示された数字が、輪郭内の判別位置（座標）を示しており、白抜きの矢印はその判別位置が白レベルのときに進む方向を、黒塗りの矢印はその判別位置が黒レベルのときに進む方向を示し、正方形で囲まれた数字は判別した数字を示している。なお、％は、それぞれ、左上端基準位置から水平方向辺長に％掛けた長さだけ水平右方向に離れた位置、垂直方向辺長に％掛けた長さだけ垂直方向下方に離れた位置、または座標を示している。例えば、基準位置から水平方向５０％、垂直方向２０％位置は、座標（５０、２０）と表示する。
【００７１】
まず、座標（５０、２０）、座標（５０、６５）、座標（１０、５０）、座標（４０、６０）、座標（２０、６０）、座標（５０、４５）を順に判定してすべての画素データレベルが白レベルの場合、その数字は「０」と判定する（Ｓ２７０１、Ｓ２７０３；白、Ｓ２７０５、Ｓ２７０７；白、Ｓ２７０９、Ｓ２７１１；白、Ｓ２７１３、Ｓ２７１５；白、Ｓ２７１７、Ｓ２７１９；白、Ｓ２７２１、Ｓ２７２３；白、Ｓ２７２５）。
【００７２】
座標（５０、２０）が黒レベルの場合（Ｓ２７０３；黒）
座標（１０、７３）が白レベルの場合は数字「７」と判別し（Ｓ２７２７、Ｓ２７２９；白、Ｓ２７３１）、黒レベルの場合は数字「４」と判別する（Ｓ２７２７、Ｓ２７２９；黒、Ｓ２７３０）。
【００７３】
座標（５０、２０）が白レベルで（Ｓ２７０３；白）、座標（５０、６５）が黒レベル（Ｓ２７０３；白、Ｓ２７０５、Ｓ２７０７；黒）の場合
座標（１０、５０）のレベルを求め、黒レベルのときは数字「９」と判定する（Ｓ２７３３、Ｓ２７３５；黒、Ｓ２７３６）。
座標（１０、５０）が白レベルで、座標（１０、９０）が白レベルのときは数字「７」と判定する（Ｓ２７３５；白、Ｓ２７３７、Ｓ２７３９；白、Ｓ２７４１）。座標（１０、５０）が白レベルで、座標（１０、９０）が黒レベルのときは数字「２」と判定する（Ｓ２７３５；白、Ｓ２７３７、Ｓ２７３９；黒、Ｓ２７４０）。
【００７４】
座標（５０、２０）が白レベル（Ｓ２７０３；白）、かつ座標（５０、６５）が白レベル（Ｓ２７０７；白）で、座標（１０、５０）が黒レベル（Ｓ２７１１；黒）の場合
座標（１０、６０）が白レベル（Ｓ２７４３；Ｓ２７４５；白）で、かつ座標（９０、２０）が白レベルのときは数字「５」と判定し（Ｓ２７４７、Ｓ２７４９；白、Ｓ２７５１）、座標（９０、２０）が黒レベルのときは数字「９」と判定する（Ｓ２７４９；黒、Ｓ２７５０）。
座標（１０、６０）が黒レベル（Ｓ２７４３；Ｓ２７４５；黒）、かつ座標（９０、３３）が白レベルのときは数字「６」と判定する（Ｓ２７５３、Ｓ２７５５；白、Ｓ２７５７）。
【００７５】
座標（９０、３３）が黒レベル（Ｓ２７５７；黒）の場合
座標（５０、４５）が黒レベルのときは数字「８」と判定する（Ｓ２７５９、Ｓ２７６１；黒、Ｓ２７６２）。
座標（５０、４５）が白レベル（Ｓ２７５９、Ｓ２７６１；白）、かつ座標（５０、６３）が黒レベルのときは数字「９」と判定する（Ｓ２７６３；Ｓ２７６５；黒、Ｓ２７６６）。
座標（５０、４５）が白レベル（Ｓ２７５９、Ｓ２７６１；白）、かつ座標（５０、６３）が白レベルのときは数字「０」と判定する（Ｓ２７６３；Ｓ２７６５；白、Ｓ２７６７）。
【００７６】
座標（５０、２０）が白レベル（Ｓ２７０３；白）、座標（５０、６５）が白レベル（Ｓ２７０７；白）、かつ座標（１０、５０）が白レベル（Ｓ２７１１；黒）の場合
座標（４０、６０）のレベルを判定し（Ｓ２７１３、Ｓ２７１５）、黒レベルのときは数字「７」と判定する（Ｓ２７１５；黒、Ｓ２７１６）。
座標（４０、６０）が白レベルのときは（Ｓ２７１５；白）、座標（２０、６０）のレベルを判定し（Ｓ２７１７、Ｓ２７１９）、黒レベルのときは数字「８」と判定する（Ｓ２７１９；黒、Ｓ２７２０）。
座標（２０、６０）が白レベルのときは（Ｓ２７１９；白）、座標（５０、４５）のレベルを判定し（Ｓ２７２１、Ｓ２７２３）、黒レベルのときは数字の「３」と判定し（Ｓ２７２３；黒、Ｓ２７２４）、白レベルのときは数字の「０」と判定する（Ｓ２７２３；白、Ｓ２７２５）。
【００７７】
以上の通り、本実施形態では、上下左右の境界位置で囲まれた切り出し範囲内の所定の座標の白黒レベルの判定を所定の順番で実行することで、切り出し範囲内の数字を判定することができる。
小さな文字は大きな文字よりも文字が細かく、すべての文字に同じ認識点を与えて認識を行った場合は、認識点の数が多く処理に時間を要するが、本発明の実施形態の場合は、切り出し範囲内の予め設定された所定箇所の画素データレベルを順番に判別し、判別順に該当する文字を判定し、確定していくので、不要な判別点の認識を省略することが可能になり、認識の確度の高い処理が可能になる。なお、本実施形態での判別点の座標は、図示の文字種、文字輪郭の場合の好ましい実施例であり、異なる文字種、文字輪郭等の場合はその文字種、輪郭等に応じて設定する。また、本発明の実施形態では、複数種の標尺に応じた座標を予めメモリに記憶しておいて、使用する標尺に応じてメモリから読み出して使用することができる。
【００７８】
【発明の効果】
以上の説明から明らかな通り本発明は、望遠鏡で視準した標尺の文字、数字を自動認識できるので、人為的な誤差を生じることがなく、より短時間でより精度の高い測量が可能になる。
【図面の簡単な説明】
【図１】 本発明の方法を適用した測量機の実施形態の概要を示すブロック図である。
【図２】 同測量機の視準視野と標尺との関係の一例を示す図である。
【図３】 同測量機の標尺の文字を自動読み取り処理の概要をフローチャートで示す図である。
【図４】 同測量機で撮像した標尺の像の一例を示す図である。
【図５】 同測量機で撮像した標尺のピッチスケールの拡大図である。
【図６】 同ピッチスケールの目盛りの様子をさらに詳細に示す図である。
【図７】 同測量機でパターンを認識するときの様子を示す図である。
【図８】 同測量機で大きな文字を認識するときの大きな文字の切り出し範囲および判別点を示す図である。
【図９】 本実施形態において「１」以外の数字を認識するアルゴリズムを模式的に示す図である。
【図１０】 同模式的に示したアルゴリズムに基づく処理をフローチャートで示す図である。
【符号の説明】
１０ ディジタルレベル
１１ 視準望遠鏡
１３ 焦点板
１５ レンズ駆動機構
１７ モータ
１９ エンコーダ
２１ エリアセンサ
２３ Ｓ／Ｈ・Ａ／Ｄ変換回路
２５ 画像処理回路
２７ フレームメモリ（ＲＡＭ）
２９ タイミングジェネレータ
３１ ＣＰＵ（演算手段、解析手段）
３３ 表示器
３５ メモリ（メモリ手段）
３７ モータドライバ
５１ 視野
５２ 鉛直ライン
５３ 水平ライン
５４ アッパスタジアライン
５５ ロアスタジアライン
１０１ 標尺
１０２ 遠距離用パターン
１０３ 大きな文字列
１０４ 小さな文字列
１０５ ピッチスケール（目盛り）[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a surveying instrument equipped with an automatic character reading device for a scale that captures an image in a telescope field of view such as a collimating telescope and recognizes the character.
[0002]
[Prior art and its problems]
In conventional surveying, an operator looks through a collimating telescope and reads characters and patterns displayed on a scale that overlap with a crosshair or the like. For example, in the case of the general-purpose measure 101 shown in FIG. 2, the interpolated values of the long distance pattern 102, the large character string 103, the small character string 104, the scale 105, and the scale 105 are read.
A digital auto level that can automatically read the scale of the standard has been developed, but automatic reading is not possible unless a standard with a code dedicated to the auto level is used, and there was no function to read characters and patterns of a general-purpose standard .
[0003]
Further, when reading characters and patterns with the human eye, it is only necessary to read the interpolated values of the small character string 104 and the pitch scale 105 at a short distance, but the pattern 102, the large character string 103, and the pitch scale 105 at a long distance. And the interpolated value must be read.
[0004]
OBJECT OF THE INVENTION
SUMMARY OF THE INVENTION An object of the present invention is to provide a surveying instrument provided with a character automatic reading device for a standard that can automatically read a character and a pattern of a standard that an operator has visually recognized and read through a telescope.
[0005]
SUMMARY OF THE INVENTION
  The present invention to achieve this objectThe scale surface of the scale with letters and black circles of a predetermined pattern located above it at predetermined intervalsSighting with a collimating telescopedo itImaging means for imaging, and the imaging meansTick surfaceStorage means for storing an image as pixel data in pixel units;Analysis for analyzing pixel data stored in the storage meansMeans, andThe analyzing means detects the upper end position of the character from the pixel data stored in the storage means, detects the level of the pixel data string in a horizontal direction from a predetermined position above the upper end position, and continuously black It is determined whether or not a level exists, and when one or more continuous black levels exist, the level of the pixel data string is detected vertically upward from the approximate center position of each of the continuous black levels. When there are a plurality of pixels, the level of the pixel data string is detected vertically upward from a substantially intermediate position between the continuous black level and the continuous black level, and the continuous black level detected by detecting the level of the pixel data string is detected. Detecting the number of bullets by number,It has the characteristics.
  The scale surface of the scale includes a pitch scale in which scales made of light and dark bands having vertical and horizontal sides are arranged in a predetermined cycle in the vertical direction, and along the pitch scale on one side across the pitch scale. The character and the black circle pattern attached at a predetermined interval, and the character smaller than the character arranged on the other side at an interval different from the interval of the character and the black circle pattern, and the analysisMeans of the scale of the pitch scaleNumber of pixels of image width or heightWhether or not is greater than or equal to a predetermined thresholdDetermine whether to recognize the smaller character or the character and black circle patternTo do.
  analysisMeans, Recognize the letters and black circle patternWhenAmong the pixel data stored in the storage means,horizontal directionAbout the pixel data stringPixel data in orderofProcessing to detect the left and right border position that contains the character by detecting the level,At different vertical positionsExecute on pixel data string, characterIs includedA boundary was detected when a pixel data string that includes left and right boundaries was detected continuously for the height of the character.VerticalTop positionAnd the lowermost pixel data string as the upper and lower positions of the character
  AndanalysisThe means is a horizontal direction at each of the detected vertical and horizontal boundary positions.And verticalDirectionPixel dataWhether there is a continuous black level in a row of pixel data for a rowDetectionHowever, when it is detected that there are continuous black levels, the detected pixel data stringExpand the detection rangeWhen it is determined that there is no continuous black level, the pixel data stringDetectionNarrow the range and correct the top and bottom and left and right border positions of the recognized character.
  Furthermore, the detected upper and lower and left and right boundariesFrom positionFind the aspect ratio of charactersThe, Whether the character is 1JudgmentTo do. When recognizing that the character is not the number 1, the pixel data level of the characteristic position corresponding to the four vertices of the substantially rectangular shape located inward from the boundary position is set.DetectionAnd other numbers depending on the combinationJudgmentTo do.
[0006]
  furtherThe analysis means includes thejudgeWhen it is determined that the character is a small one-digit to three-digit number, the approximate range of the character of each digit is set on the basis of the predetermined position on the scale of the pitch scale, and the upper, lower, left and right sides surrounding the approximate range of each character Check the pixel data level ofDetectFind the aspect ratio of the character based on the boundary position, and whether the character is a number 1JudgmentTo do.
  The analyzing means determines that the character is not the number 1.JudgmentWhen calculating, one feature position located inward from the boundary position is calculated.Feature locationProcessing that checks the pixel data level, uses the possible number as a result of the determination, calculates another feature position based on this determination result, and similarly checks the pixel data levelA possible numberIs oneRepeat until.
[0007]
DETAILED DESCRIPTION OF THE INVENTION
The present invention will be described below with reference to the drawings. First, an outline of an optical system according to an embodiment of a surveying instrument (digital level) to which the present invention is applied will be described with reference to FIG. 1 and an example of general-purpose staff with reference to FIG.
[0008]
This digital level 10 serves as a collimating telescope 11 from the collimating object side, from the collimator side, an objective lens group (objective optical system) L1, a focus adjusting lens group (focus adjusting optical system) L2, and a compensation / upright optical element (compensation / upright) An optical system) L3, a beam splitting optical element (dividing optical system) L4, a focusing screen 13, and an eyepiece lens (eyepiece optical system) L5. The focusing screen 13 is provided with cross lines (vertical lines and horizontal lines) and stadia lines (uppasta lines and roaster lines). Therefore, when the operator collimates the scale with the collimating telescope 11 through the eyepiece lens group L5, the scale surface of the scale formed on the focusing screen 13 through the objective lens group L1 and the focusing lens group L2. And the cross lines (vertical lines and horizontal lines) and stadia lines (appastia lines and roaster lines) formed on the focusing screen 13 are visually recognized. The operator visually observes the image of the scale surface on which the vertical line, the horizontal line, the Appastasia line and the Roasteria line overlap, and reads the scale pattern, numbers, scale, etc. to obtain the survey value (see FIG. 4). . As the compensation / upright optical element L3, for example, a suspended compensator erecting prism is used.
[0009]
The digital level 10 includes an area sensor 21 as an imaging unit that captures an image of the scale surface of the scale formed by the objective lens group L1 and the focus adjustment lens group L2. The light beam incident on the light beam splitting optical element L4 is split into a light beam that passes through the split surface L4D and travels toward the eyepiece lens L5, and a light beam that is reflected by the split surface L4D and travels in the direction of the area sensor 21. The light beam divided in the direction of the area sensor 21 is incident on the light receiving surface 21 a of the area sensor 21. Since the light receiving surface 21a of the area sensor 21 is disposed at an equivalent position to the focal plane of the focusing screen 13, a clear image of the scale surface is formed on the light receiving surface 21a when the scale surface of the scale is in focus. The
[0010]
Although not shown in detail, the digital level 10 includes an automatic focus adjustment device that detects a focus state by a contrast method using image data captured by the area sensor 21 and adjusts the focus. In this configuration, the focus adjustment lens unit L2 is moved to the in-focus position where the contrast of the image of the scale imaged by the area sensor 21 is highest. In addition, the automatic focus adjustment device includes a lens drive mechanism 15 that moves in the optical axis direction while holding the focus adjustment lens group L2, a motor 17 as a power source that drives the lens drive mechanism 15, and the focus adjustment lens group L2. An encoder 19 for detecting the position is provided.
[0011]
The main circuit configuration of the digital surveying system of the digital level 10 will be described with reference to FIG. The light receiving surface 21 a of the area sensor 21 is disposed at a position equivalent to the focusing screen 13. That is, when collimating the scale with the collimating telescope 11, an image of the scale surface of the scale is formed on the light receiving surface 21 a of the area sensor 21.
The light receiving surface 21a (sensor region) of the area sensor 21 is generally rectangular, and the image in the longitudinal direction of the scale when the area of the light receiving surface 21a is the horizontal direction and the direction of the narrow area is the vertical direction. When the area sensor 21 is mounted so that an image is formed in the vertical direction of the area sensor 21, the vertical direction of the light receiving surface 21a coincides with the vertical direction in the collimating visual field in a state where the collimating telescope 11 is installed horizontally. .
Further, when the area sensor 21 is attached so that the longitudinal image of the scale is formed in the horizontal direction of the area sensor 21, the horizontal direction of the light receiving surface 21a can be viewed in a state where the collimating telescope 11 is installed horizontally. It coincides with the vertical direction in the near field of view.
[0012]
The area sensor 21 converts the image formed on the light receiving surface 21a into an electrical pixel signal by each light receiving element (photoelectric conversion element), accumulates it, and outputs it in units of pixels. As the area sensor 21, a so-called CCD image sensor, a CMOS image sensor, or the like can be used. In the present embodiment, an image captured by the area sensor 21 is framed as pixel data in units of pixels via the image processing circuit 25. It is stored in a memory (RAM) 27. That is, the pixel signal output from the area sensor 21 is quantized (digitized) by the S / H (sample hold) / A / D conversion circuit 23. The quantized pixel signal is stored in a frame memory (RAM) 27 via the image processing circuit 25.
[0013]
The digital image data for one frame written in the frame memory 27 is read into the image processing circuit 25 in units of pixel data, subjected to predetermined processing, and predetermined image analysis is performed by the CPU 31 serving as analysis means. The CPU 31 analyzes the image information of the scale surface of the scale taken by the area sensor 21 by the image analysis process, identifies and calculates the level and distance. Then, the CPU 31 displays data on the obtained level or distance on the display 33 to notify the operator. The frame memory 27 is also used as a work area for image analysis.
[0014]
Imaging operation of the area sensor 21, for example, sweeping out unnecessary charges, accumulation of charges (imaging), output of accumulated charges (pixel signal output), sampling and A / D conversion operation of the S / H / A / D conversion circuit 23, image A series of operations such as the operation of the processing circuit 25 is executed in synchronization with the clock pulse output from the timing generator 29.
[0015]
The CPU 31 displays items, parameters, and the like necessary for the surveying on the display 33 before the surveying process, and displays the survey results and the like on the display 33 when the automatic surveying process is executed. In the automatic focus adjustment mode, the CPU 31 detects the focus state by the contrast method for detecting the contrast of the standard image captured by the area sensor 21, and moves the focus adjustment lens group L2 from the focus state to the in-focus position. The motor 17 is driven via the motor driver 37 to obtain the necessary driving direction and driving amount, and the focus adjustment lens group L2 is driven to the in-focus position via the lens driving mechanism 15.
[0016]
Further, the position of the focus adjustment lens unit L2 in the optical axis direction is detected via the encoder 19. Since the relationship between the position of the focusing lens group L2 and the object distance to be focused is uniquely determined, if the position of the focusing lens group L2 can be detected, the approximate distance from the position to the focused object can be obtained.
[0017]
Note that selection and input of parameters necessary for automatic surveying are performed by an input means, for example, the keyboard 39. The keyboard 39 can be provided with, for example, a numeric keypad, a function key, or a dedicated operation key.
[0018]
In the memory 35, correction information necessary for normal surveying such as an adjustment reference value and a temperature compensation coefficient, and data necessary for analyzing and identifying a scale scale pattern, number, or scale of the scale are stored for a plurality of scales. Prepared and written.
[0019]
As the memory 35, an EEPROM or a removable nonvolatile memory is suitable. EEPROM, flash ROM, etc. are electrically rewritable memories that can be stored without losing their contents even when the power is turned off. It is suitable for storing different adjustment values and setting values for each apparatus. It is advantageous for a surveying instrument that needs to change the adjustment value periodically for precise measurement.
Especially in such an apparatus, the relationship between the horizontal line when looking through the telescope and the horizontal line set on the light receiving surface of the imaging means is important. If this does not match, the adjustment value is rewritten and the adjustment value is rewritten. The non-volatile memory is suitable because it is necessary.
[0020]
The above digital surveying, that is, the automatic reading process of the level (the reading of the scale on the horizontal hairline) and the horizontal distance is selected by the CPU 31 by the operator in advance through the keyboard 39 when the digital surveying start key is turned on. Data corresponding to the measured scale code information is read from the memory 35 and executed. Before performing the automatic reading process of the level and horizontal distance, in order to enable accurate measurement, the object to be measured is either focused in advance by an automatic focusing device or manually adjusted by the operator. Keep it in focus.
[0021]
The visual field 51 of the collimating telescope 11 is regulated by the field stop of the collimating telescope 11, and usually has a circular outline as shown in FIG. Within this field of view 51, a vertical line 52 and a horizontal line 53 as crosshairs, and an appastia line 54 and a roasteria line 55 as stadia can be visually recognized.
On the other hand, the effective light receiving surface 21a of the area sensor 21 does not need to be in the same area as the visual field 51, but in this embodiment, includes an area above the upper pasta line 54 and an area below the roasteria line 55. It is set.
[0022]
  In this embodiment, the light beam splitting optical element(prism)Since L4 is disposed closer to the objective lens group L1 than the focusing screen 13, when the collimating telescope 11 is viewed from the eyepiece lens L5, it is placed on the object-side image.crossline(Lines 52, 53, 54, 55)Appear to overlap. HoweverThe object side image is formed on the light receiving surface 21a of the area sensor 21, but the images of the lines 52, 53, 54, and 55 are not formed. Further, the positions of the images of the lines 52, 53, 54, and 55 that overlap the image on the object side are always the same position in the field of view even if the observation target changes. Therefore, the position (coordinates) on the area sensor 21 corresponding to the images of the lines 52, 53, 54, and 55 is obtained in advance and stored in the memory 35.
  Here, the positions of the images of the lines 52, 53, 54, and 55 are stored as pixel addresses on the area sensor 21, but the images of the lines 52, 53, 54, and 55 do not necessarily overlap the light receiving element. The image positions of the lines 52, 53, 54, and 55 are obtained as interpolation values together with pixel addresses, and the interpolation values are also stored in the memory 35.
[0023]
Further, in consideration of the inclination of the area sensor 21, addresses of two points at least separated from one line 52, 53, 54, 55 are obtained. For example, a crosshair chart is prepared, and this is accurately set in the vertical direction. The crosshair chart is collimated through the collimating telescope 11, and the lines 52 and 53 and the crosshairs are made coincident. At this time, the peak of the line portion is detected from the image data of the cross line formed on the area sensor 21 and its address (vertical position and horizontal position) is calculated. However, since the crosshair chart at this time is a reference for horizontal and vertical, the chart position is set and adjusted accurately, and the vertical and horizontal lines 52 and 53 are adjusted as accurately as possible. To do.
[0024]
Next, the configuration of a general-purpose staff (standard measure) that can be automatically recognized by the digital level 10 and is widely used for surveying will be described. There are various types (grades, materials) and various displays (character strings, patterns, scales) as general-purpose staffs (staff). As an example, the configuration of the most general-purpose standard scale 101 will be described. . The standard 101 is an aluminum three-drawer box measure that can measure up to 5m, with the width of the bottom step being 50mm, the width of the middle step being 45mm, and the width of the narrowest step being 40mm. is there.
[0025]
FIG. 2 shows the scale surface (display surface) of the staff 101 observed in the visual field 51 of the collimating telescope 11. The scale display of the scale 101 includes a long-distance pattern 102, a long-distance character string 103, a short-distance character string 104, and a pitch scale 105 as a 10-mm pitch scale. Normally, the ground color of the scale surface of the staff 101 is white, the long distance pattern 102, the character strings 103 and 104, and the pitch scale 105 are black or red.
[0026]
The character height of each number in the large character string 103 is 50 mm, and the character height of each number in the small character string 104 is 6 mm. The upper side of each number of the character strings 103 and 104 and the upper side of the scale (black belt) of the pitch scale 105 are set so as to be positioned on the same horizontal line when the scale 101 is installed vertically. The height above the upper side is the height indicated by the number.
The character width of large characters is 11 mm, 9 mm, and 8 mm in order from the bottom, and the character width of small characters is 2.5 mm regardless of the step.
[0027]
The long-distance pattern 102 is a black circle attached to the upper part of each number of the large character string 103, and one black circle indicates a 1 × 1000 mm (1 m) level. For example, in FIG. 2, three black circles indicate a 3 × 1000 mm (3 m) level, and two black circles indicate a 2 × 1000 mm level. The large character string 103 is composed of a single-digit number, and its unit is 100 mm (10 cm). For example, the number “2” of the large character string 103 in FIG. 2 indicates that 2 × 100 = 200 mm. In the case of FIG. 2, three black circles are attached as the long distance pattern 102 on the upper part of the character “2” of the large character string 103, and therefore the number 2 of the large character string 103 indicates the 3000 + 200 = 3200 mm range. I understand that.
[0028]
The small character string 104 is indicated by a three-digit number. The smallest unit (the first digit) is 1 cm (10 mm), the second digit is 10 cm (100 mm), the third digit is 100 cm (1,000 mm or 1 m). The table is displayed. For example, in FIG. 2, the three-digit number “323” in the small character string 104 indicates 323 cm, 3230 mm, or 3 m23 cm.
[0029]
The 10 mm pitch scale 105 is composed of two rows, the left 5 mm scale S1 is a repetition of a 10 mm period light and dark band formed in 5 mm units, and the right 10 mm scale S2 is formed in 10 mm units. Repeated light and dark bands with a period of 20 mm. Note that in the case of surveying by visual inspection, normally, at a short distance, the white or black band of the pitch scale 105 is read to 1/5 of the width of 5 mm, that is, 1 mm, but according to the automatic reading process of the embodiment of the present invention, A value smaller than 1 mm can be read by a predetermined fraction calculation.
[0030]
Further, in the case of visual leveling, the operator can see the level (level height) in the field of view of the collimating telescope 11 and the distance pattern 102 and the character strings 103 and 104 on the scale surface of the scale 101 where the horizontal line 53 overlaps. And the pitch scale 105 are substituted into a predetermined mathematical formula for calculation. The distance is obtained by reading the distance pattern 102 on the scale surface of the scale 101 where the upper and lower stadia lines 54 and 55 overlap, the numbers of the character strings 103 and 104, and the pitch scale 105, and obtaining the length between the stadia lines 54 and 55. Calculation is performed by substituting the obtained length into a predetermined mathematical expression. In the figure, reference numeral 52 denotes a crosshair vertical line, which is set so that the collimation axis (optical axis) O of the collimating telescope 11 passes through the intersection of the vertical line 52 and the horizontal line 53.
[0031]
An embodiment of the present invention for automatically recognizing a scale character collimated by the digital level 10 will be described below with reference to FIGS.
This character / pattern recognition process is a process in which the CPU 31 reads and executes a program stored in the ROM 35 when an operation switch is pressed by an operator or the like. In order to facilitate understanding, the measure 101 is used as a general-purpose measure. Before entering this process, the collimating telescope 101 collimates the staff 101 and drives the focus adjustment lens group L2 to focus on the staff 101 to drive the area sensor 21 for one frame. Are written in the frame memory 27 as pixel data.
In this embodiment, when the collimating telescope 11 is focused on the scale 101 and the scale in the range of 105 cm to 112 cm on the scale 101 is collimated, that is, the scale shown in FIG. 4 is included in the visual field. This will be described as an example.
[0032]
Before entering the character / pattern recognition process, first, the used scale is determined, and the scale portion is detected. Thereafter, the character / pattern recognition process is started.
[0033]
In the character / pattern recognition process, first, the type of character to be recognized is determined (S21). In this embodiment, it is determined whether the character string 103 is large or the character string 104 is small.
When it is determined that a large character is recognized, the position of the large character is detected (S22), the large character is recognized based on the detected position (S23), and the pattern attached to the upper portion of the recognized large character. Is recognized (S24). Then, from the number of scales, that is, the position represented by the recognized large character, the number of 10 mm scales between the position of the horizontal line to be obtained is obtained (S25).
When it is determined that the character is recognized as a small character, the position of the small character is detected (S26), the recognition process for each three-digit small character is performed (S27), and the recognition process is terminated.
[0034]
Details of the above character / pattern recognition processing will be further described with reference to FIGS. In the following description, the image of the scale 101 formed on the light receiving surface 21a of the area sensor 21 and the coordinates on the light receiving surface 21a will be described as a reference.
[0035]
S21: Determination of character string to be recognized
Whether the large character string 103 or the small character string 104 is recognized is determined based on an image formed on the light receiving surface 21 a of the area sensor 21. Normally, a small character string is set to be recognized in the short distance range, and a large character string is set to be recognized in the long distance range.
In the case of the present embodiment, whether the range is a short range or a long range is determined by detecting the image width or image height (horizontal length or vertical length) of the scale of the pitch scale 105 from the pixel data of the captured image. To do.
[0036]
(1-1) In the case of the telescopic staff 101, patterns, characters, and scales are usually drawn in black or red on a white background. Therefore, a portion of the scale whose luminance is low, that is, a portion drawn in black or red (hereinafter referred to as “black portion”) is detected, and an arbitrary horizontal side (any one of H1 to H3) and a vertical direction The ratio of the lengths of the two sides (any one of V1 to V4) is obtained to identify the step number.
[0037]
An embodiment of a method for obtaining the side length will be described with reference to FIG.
(1-1-1) Read a line of pixel data in a vertical direction at a predetermined horizontal direction detection start position, and obtain a simple average value. In normal collimation surveying, the horizontal detection start position is usually set so that the intersection of the crosshairs coincides with the scale in the horizontal center of the scale when collimating the standard with a collimating telescope. The center position corresponds to the horizontal direction detection start position. However, since the scale portion may not be located near the center of the scale depending on the scale, in such a case, a position shifted to the left and right from the center is set as the horizontal direction detection start position.
The pixel data to be used is performed by using all the data in the vertical direction at the horizontal direction detection start position determined by the scale or only a part of the data.
However, when the measurement is performed near the upper end of the scale, there is a possibility that the portion above the horizontal line 53 may be lost. In this case, only pixel data below the horizontal line 53 is used.
[0038]
In the case of a monochrome area sensor, data for one column of pixels is read. In the case of a color area sensor, data for two columns of pixels is read. However, when the color area sensor is used for detection using only a specific color component, one line is sufficient. For example, when the color area sensor includes R, G, and B primary color filters and uses only the G component, only the pixel data of the pixel column to which the G filter is attached is used.
[0039]
(1-1-2) Each pixel data is binarized by comparing the average value of the obtained pixel data for one column with each pixel data one by one. Thereby, bright and dark pixel data is obtained.
[0040]
(1-1-3) The number of consecutive pixel data of the same level is sequentially obtained from the binarized pixel data. In the illustrated embodiment, since the levels of the pixel data above and below the pixel data at the upper and lower ends are unknown, pixel data continuous at the same level as the pixel data at the upper and lower ends is not counted.
In the area of FIG. 4B, in order from the bottom, white level; 5, black level; 5, white level; 5, black level; 6, white level;
In the area of FIG. 4C, in order from the bottom, white level; 3, black level; 3, white level; 4, black level; 3, white level; 1, black level; 5, white level; Level; 2, white level; 2,.
[0041]
(1-1-4) The number of continuous pixel data at the same level (white or black level) and the number of continuous pixel data at the same level (white level is white level, black level is black) Find the absolute value of the difference between level and).
In the area of FIG. 4B, white-white = 0, black-black = 1, and white-white = 0.
In the region of FIG. 4C, white-white = 1, black-black = 0, white-white = 3, black-black = 2, white-white = 4, black-black = 3, white-white = 3.
[0042]
(1-1-5) Obtain the average value of the differences between the number of consecutive pixel data obtained at the same level.
In the area of FIG. 4B, it becomes 0.33, and in the area of FIG. 4C, it becomes 2.28.
[0043]
(1-1-6) If the average value of the differences obtained in (1-1-5) is less than or equal to a predetermined threshold value, the column is judged as a scale part. Conversely, if the average difference is not less than or equal to the predetermined threshold value, the column is determined not to be a scale portion.
Here, for example, when the predetermined threshold value is 1, FIG. 4B is smaller than the threshold value, so it is determined as a scale portion, and FIG. 4C is larger than the threshold value and is not a scale portion. Judge.
[0044]
(1-1-7) When it is determined that it is a scale, move the column to be analyzed one pixel or two pixels to the left or right from (1-1-1) to (1-1-6 ) Is repeated until a column that is determined not to be a scale is detected.
If it is determined from the first detection result that it is not a scale portion, the processing from (1-1-1) to (1-1-6) is performed one pixel on the left or right until the scale portion is detected. Repeat while moving for minutes or more.
[0045]
(1-1-8) Obtain the average value per column of the number of consecutive pixel data of the same level obtained in (1-1-3), and change the pitch scale between columns where it changes greatly in adjacent columns. The boundary between the 105 millimeter scale S1 and the 10 millimeter scale S2 is 105 (FIG. 2).
[0046]
The lengths of the sides H1 to H3 in the horizontal direction can be obtained from the intervals between the respective positions because both ends of the scales S1 and S2 and the boundary positions are obtained.
The vertical directions V1 to V4 can be obtained by obtaining the number of pixel data in which the same level of the scales S1 and S2 continues within the respective ranges.
[0047]
The length of the side in the vertical direction is the same regardless of the level because it is a scale, but the length of the side in the horizontal direction is different for each level in the case of the illustrated measure 101. Therefore, if the ratio of the length of the side in the horizontal direction to the length of the side in the vertical direction is obtained, it is possible to determine which level is collimated.
[0048]
  (1-2) If a black part is detected, the length of any one side of the black part (H1-H3, V1-V4), and (the aboveHow many levels are you looking atJudgmentBased on the result, which character string is recognized, that is, which one of the large character string 103 or the small character string 104 is recognized.JudgmentTo do.
[0049]
Even if it is a general-purpose measure, the character of the measure is different in size, shape, arrangement, etc. depending on the type of measure, so there are multiple built-in parameters and algorithms corresponding to those differences, and the measure used from among them Select and use parameters and algorithms suitable for the
[0050]
S22: When the length of an arbitrary side is smaller than a predetermined threshold (position detection of a large character string 103)
If the detected length of the vertical or horizontal side is smaller than the predetermined threshold value, it is determined that the large character string 103 is recognized. That is, in this case, since the staff 101 is installed at a long distance, the character image of the small character string 104 is small and the number of pixels occupied by each character is too small. Hereinafter, the recognition of the large character string 103 will be described with reference to FIG. In the present embodiment, the pixel data level is checked in the region on the left side of the pitch scale 105, and a horizontal column in which black levels are continuous is detected.
[0051]
  (2-1) Located on the left side of the scale S1 of the pitch scale 105Black circleIn the region where P and the large character C1 exist(Fig. 5)The level of pixel data in a horizontal rowDetectionIs executed at intervals of 10 mm in the vertical direction to detect a portion where a plurality of black levels continue.
  In this way, by repeating the check at one graduation interval at a plurality of positions, a large character C1 between the upper and lower columns where a plurality of black levels no longer appear continuously in the horizontal direction andBlack circleThe presence of P can be detected.
  The position where the black circle P is detected is the black circle.P and sizeNaletterC1When detecting the range that existsFor horizontal line 53It is obtained from the positions of the upper and lower sides of one scale obtained near the position, and the other detection positions are obtained by adding / subtracting the number of pixels corresponding to the scale period from that position.
[0052]
(2-2) When a row with consecutive black levels is detected continuously in the vertical direction by the height of the character, the highest detection position where the black level is detected is the pattern recognition start position, the second from the top. The upper end of the scale including the th detection position is the upper end position of the character, and the lower end of the scale including the lowermost detection position is the lower end position of the character. However, if the number of detected black levels is only the size of the character, it is determined that the portion where there is no pattern, that is, the area below 1 m of the scale is being measured.
(2-3) Furthermore, when a black level continuous in the horizontal direction is detected, the position at which the white level is first changed to the black level and the position at which the black level is finally changed to the white level are written to the memory 27. deep. The space between these two level change positions corresponds to the horizontal line width of the character.
(2-4) In the result of checking the pixel data level of the horizontal column in (2-3), the right edge of the character first changes from white to black level in the column where continuous black level is detected. As for the left edge of the pixel and the character to be used, the pixel that finally changes from black to white level is set as the position of the right edge and the left edge, respectively. That is, the positions of four sides and corners (vertices) of the rectangle surrounding the character are obtained.
(2-5) Further, with respect to the obtained vertical and horizontal edge columns, it is confirmed whether there is a continuous black level portion in one column of pixel data. If there is no continuous black level, move the pixel data to be checked inward (downward at the upper end, upward at the lower end, rightward at the left end, leftward at the right end), If there is a part where the black level continues, move to the outside (upward at the upper end, downward at the lower end, leftward at the left end, rightward at the right end) Modify the part. That is, the boundary between the line where the character line is broken and the line where the line is not broken is detected. Among the boundaries detected in this way, the column where the line of the character is not broken is set as the boundary position of the top and bottom and the left and right of the character.
[0053]
As described above, it is possible to achieve both shortening of detection time and accurate cut-out by first obtaining the rough position of the area including the character and narrowing the range of the character from the obtained rough boundary position. .
[0054]
S23: Recognize large characters (determine hundreds of digits)
(3-1) The vertical / horizontal ratio of the large character is obtained from the vertical / horizontal boundary position information of the character obtained in S22, and whether the number is “1” or any other number is recognized from the result.
(3-2) If “1” is not determined, feature points determined at a fixed ratio within the cut-out character range, for example, four characters A, B, C, and D in the case of a character as shown in FIG. Characters can be recognized simply by checking the level of dots. Table 1 shows the relationship between the levels of the points A, B, C, and D and the numbers.
[0055]
[Table 1]
    A B C D Result (number)
    White black black black 2
    White black white black 3
    White white black black 4
    Black Black White White V (5)
    Black white black black 6
    White black whiteWhite         7
    Black black black black 8
    Black black white black 9
[0056]
(3-3) In the case of this character profile (font), 2-9 characters can be identified by recognizing 4 black and white levels, but if the number of feature points is increased according to the method of (3-2), Accuracy is improved.
(3-4) When judging the level of a feature point using not only one pixel point data corresponding to each feature but also using pixel data of a plurality of neighboring points, the effect of level fluctuation due to noise, etc. is eliminated, Misjudgments are reduced.
[0057]
As described above, according to the character recognition method of the present embodiment, it is possible to identify and recognize numbers by confirming the level of pixel data of a small number of feature points (four points), so that the recognition speed can be improved. In addition, since the characters of the general-purpose staff are thick and have a geometric shape, high recognition accuracy can be maintained even when four points are recognized as in this embodiment.
[0058]
S24: Pattern102 (black circle P)Recognizing (thousand mm, seeking the place of a meter)
  (4-1) The pattern 102 exists within a certain distance from the position of the large number obtained in S23 and the scale portion, and the position is on the scale from the position of the upper end of the large character string 103 and the left end of the scale. Then it exists at a certain distance. As a method for determining the range, the interval from the range where the largest pattern exists to the upper end of the large character and the left end of the scale is stored in the memory 35 as a ratio to the scale period, and the scale period When the number of pixels included is calculated and these ratios are multiplied, the interval from each position can be obtained.
[0059]
  (4-2)Black circle PThe first column to check the data level changes necessary to recognize(Horizontal pixel data string)To decide. The first column is from the range obtained in (4-1)OutsideTo be the start and end pointsTheMake sure that the first and last level change points can be detected. In the case of this staff 101, large charactersScale S1 with a period of 10mm from the top edgeMake sure to place a pattern at a position one cycle above102 (black circle P)Because it is arranged so that the circular center of(Fig. 5)By first checking the level change in this column,102 (black circle P)The presence or absence of can be confirmed reliably.
  All vertical scalesPixelThe average value is obtained from the data, and the average value is used as a threshold for binarization. And this columnPixelFor data, from right to leftThresholdCompared with, repeat the determination of white or black level,When the determination of the black level is two or more consecutive (continuous black level), it is black.
  (4-3) EachHorizontal or vertical columnEach time the pixel data level changes from white to black, a change point that changes to white is detected, and the number of change points is counted.
[0060]
  (4-4) Recognize patterns in the following steps based on the number of change points in each horizontal and vertical row.
  (4-4-1)HorizontallyWhen there was no change point,There are 0 black circlesIt is determined that it is 0 m (less than 1 m) (FIG. 8A)Circled number 1).
  (4-4-2)HorizontallyWhen there is a set of change points that change from white to black to white,One black circleDetermined to be 1 meter (1 meter or more and less than 2 meters) (FIG. 8B)Circled number 1).
  (4-4-3)HorizontallyWhen there are two sets of change points that change from white to black to white(Figure 1 (C) circled number 1)Change from white to black in the first pairChange point toThen change from black to whiteChanging pointApproximate center coordinates of(Approximate middle position of black circle)From vertical direction(Vertically upward)Next, it is determined in turn whether the pixel data on the vertical column is black level or white level.
  (4-4-3-1)VerticallyWhen there is a set of change points that change from black to white and then change from white to black to white,4 black circles,Determined to be 4m (4m or more and less than 5m) (Fig. 8C)Number 2 with circle).
  (4-4-3-2)VerticallyIf there is no change point from white to black to white after changing from black to white(Figure 2 (D) circled number 2), HorizontallyAt firstBlack to whiteChange point to change,afterwardsWhite to blackChange toThe approximate center of the two change pointsCoordinates (substantially intermediate position between black circle and black circle)From vertical directionVertical (vertically upward)Pixel data on the columnIs black level or white levelIn order(Number 3 in FIG. 8D).
  (4-4-3-3)VerticallyWhen there is a set of transition points from white to black and from black to white3 black circlesIt is determined that it is 3m level (3m or more and less than 4m) (in Fig. 8D)Circled number 3).
  (4-4-3-4)VerticallyWhen there is no transition point from white to black or from black to white2 black circlesIt is determined to be 2m units (2m or more and less than 3m) (Fig. 8E)Circled number 3).
  Changes in black and white in the horizontal and vertical rows (number of consecutive black levels)And patternHeight represented by (number of black circles) (numerical value)Table 2 shows the relationship.
[0061]
[Table 2]

[0062]
In the case of a scale that is displayed as a numerical value instead of the pattern 102, the processing is performed in the same manner as the recognition of small characters in S27 to recognize the numerical value.
[0063]
Regarding the pattern 102, when the existence range is determined, even if the existence range is determined from the position of the large character and the scale, there is a possibility that some deviation occurs. As a method of recognizing a pattern, a method of judging at the level of a feature point determined by a certain ratio within the existence range is conceivable, as in the case of character recognition. There is sex. Considering the modification of the pattern existence range to prevent this misrecognition, the data level within the pattern existence range is confirmed, the point where the data level changes is detected, and they are located on the outermost side By determining the portion as the end of the existence range, the existence range of the pattern can be corrected. However, when the pattern existence range is corrected by such a method, since the dimensions of the outer shape of the pattern are different from each other, the feature points are further shifted, and the possibility of erroneous recognition increases.
Therefore, in the present embodiment, it is possible to increase the recognition rate by first confirming and discriminating how to change the level of a column in which the presence of a pattern can be surely confirmed from data.
[0064]
S25: Calculate the number of scales (determine 10mm)
(5-1) Calculate the number of pixels included between the upper side UC of the large number and the position to be obtained (horizontal line 53, upstadia line 54, or roasteria line 55).
(5-2) Divide the obtained number of pixels by the number of pixels included in one cycle of the 10 mm scale, and extract only the integer part to obtain the number of scales.
[0065]
To find the number of scales from the position of a large number to the position to be measured, it is only necessary to count the number of scales, but in order to handle this simply, the vertical coordinate of the upper side of the large character is set to the same vertical position as the initial position. While checking the pixel data on the direction column in order, detect the change points that change from white to black and black to white up to the measurement position, and how many change points for one scale exist in the measurement range. You only have to count.
[0066]
However, in this method, it is necessary to detect the change point by comparing the levels of the previous pixel data and the current pixel data in all the pixel data at an interval to be measured.
On the other hand, when the position information (coordinates) on the area sensor of the upper end position of the character and the position to be measured is taken as the difference between the coordinates of the positions, the number of pixels included between the two positions is obtained. Dividing the number of pixels by the number of pixels included in one period and extracting only the integer part of the quotient, the number of scales can be obtained. Unless the upper end of the character and the desired position are extremely close, this process takes less time.
[0067]
S26: When the length of an arbitrary side is larger than a predetermined threshold (the position of the character in the small character string 104 is detected)
(6-1) Find the range where small characters exist based on the right end position of the scale and the horizontal and vertical lengths of the scale.
The length from the right end position of the scale to the small character relative to the horizontal length of the scale is memorized in advance, the number of pixels included in the horizontal side of the scale is counted, and the ratio of each pixel is multiplied by that number. Find the horizontal position (right and left edges of the character).
In the vertical direction, the upper end of the black level of the scale coincides with the upper end of the character, and the lower end can be obtained by multiplying the number of pixels corresponding to one cycle of the scale.
(6-2) For the end positions obtained in (6-1), check the data corresponding to each of the four sides, and detect two columns where the columns where the black level exists continuously and the columns where the black level does not exist are adjacent. A column in which black levels continuously exist at this time is defined as the end of the character.
The end of the character is the position of each side of the quadrangle enclosed so that the small character is inscribed, and the detection of the small character is to obtain all the positions of these four sides.
[0068]
In such a small character detection process, in the same way as the detection of the position of a large character, a rough position is first obtained, and then the range of the character is narrowed down to achieve both shortening of detection time and accurate cutout. Yes.
[0069]
S27 Three-digit small character recognition (recognizes numbers in the 1,000mm, 100mm, and 10mm positions)
(7-1) The aspect ratio of the small character is obtained from the position information of the small character obtained in S26, and whether the character is “1” or any other number is recognized from the result. The aspect ratio is about 5: 1 in the case of “1”, and about 3: 1 in other cases. Therefore, if the aspect ratio is larger than 4: 1 as a reference, it is determined as 1. If it is less than that, it is determined that it is other than “1”.
(7-2) If it is not recognized as “1” in (7-1), the pixel data at the position where the numerical value can be reliably recognized as shown in FIG. To recognize. That is, each time the level of pixel data at each position is confirmed, the character is recognized by a method of determining a number determined by the combination of the information.
[0070]
In the present embodiment, when the aspect ratio is 4: 1 or less, the upper left end of a small character is recognized as a reference position and recognized in the following order. 9 and 10 are a schematic diagram and a flowchart showing an algorithm for recognizing numbers other than “1”. In FIG. 9, a vertically long rectangle is a cutout range (contour) of a small character, a number indicated by% indicates a discrimination position (coordinate) within the outline, and a white arrow indicates that the discrimination position is white. The direction of progress when the level is reached, the black arrow indicates the direction of advance when the determination position is the black level, and the numbers enclosed in squares indicate the determined numbers. Note that% is a position that is horizontally separated from the upper left reference position by the length multiplied by% to the horizontal side length, a position that is vertically separated by a length multiplied by% to the vertical side length, Or the coordinates are shown. For example, a position of 50% in the horizontal direction and 20% in the vertical direction from the reference position is displayed as coordinates (50, 20).
[0071]
First, coordinate (50, 20), coordinate (50, 65), coordinate (10, 50), coordinate (40, 60), coordinate (20, 60), coordinate (50, 45) are determined in order and all When the pixel data level is the white level, the number is determined to be “0” (S2701, S2703; white, S2705, S2707; white, S2709, S2711; white, S2713, S2715; white, S2717, S2719; white, S2721) S2723; white, S2725).
[0072]
When the coordinates (50, 20) are at the black level (S2703; black)
When the coordinate (10, 73) is at the white level, it is determined as the number “7” (S2727, S2729; white, S2731), and when it is at the black level, it is determined as the number “4” (S2727, S2729; black, S2730). .
[0073]
When coordinates (50, 20) are at the white level (S2703; white) and coordinates (50, 65) are at the black level (S2703; white, S2705, S2707; black)
The level of the coordinates (10, 50) is obtained, and when it is the black level, it is determined that the number is “9” (S2733, S2735; black, S2736).
When the coordinates (10, 50) are at the white level and the coordinates (10, 90) are at the white level, the number “7” is determined (S2735; white, S2737, S2739; white, S2741). When the coordinates (10, 50) are at the white level and the coordinates (10, 90) are at the black level, the number “2” is determined (S2735; white, S2737, S2739; black, S2740).
[0074]
When coordinates (50, 20) are white level (S2703; white), coordinates (50, 65) are white level (S2707; white), and coordinates (10, 50) are black level (S2711; black)
When the coordinate (10, 60) is the white level (S2743; S2745; white) and the coordinate (90, 20) is the white level, it is determined that the number is “5” (S2747, S2749; white, S2751). When 90, 20) is at the black level, it is determined that the number is “9” (S2749; black, S2750).
When the coordinates (10, 60) are the black level (S2743; S2745; black) and the coordinates (90, 33) are the white level, the number “6” is determined (S2753, S2755; white, S2757).
[0075]
When the coordinates (90, 33) are at the black level (S2757; black)
When the coordinates (50, 45) are at the black level, it is determined that the number is “8” (S2759, S2761; black, S2762).
When the coordinates (50, 45) are the white level (S2759, S2761; white) and the coordinates (50, 63) are the black level, the number “9” is determined (S2763; S2765; black, S2766).
When the coordinates (50, 45) are the white level (S2759, S2761; white) and the coordinates (50, 63) are the white level, the number “0” is determined (S2763; S2765; white, S2767).
[0076]
When coordinates (50, 20) are white level (S2703; white), coordinates (50, 65) are white level (S2707; white), and coordinates (10, 50) are white level (S2711; black)
The level of the coordinates (40, 60) is determined (S2713, S2715), and when the level is black, it is determined as the number “7” (S2715; black, S2716).
When the coordinates (40, 60) are at the white level (S2715; white), the level of the coordinates (20, 60) is determined (S2717, S2719), and when the coordinates are at the black level, the number “8” is determined (S2719; Black, S2720).
When the coordinates (20, 60) are at the white level (S2719; white), the level of the coordinates (50, 45) is determined (S2721, S2723), and when the coordinate is at the black level, the number is determined to be “3” (S2723). Black, S2724), and white level, it is determined that the number is “0” (S2723; white, S2725).
[0077]
As described above, in the present embodiment, it is possible to determine the numbers in the cutout range by executing the black and white level determination of the predetermined coordinates in the cutout range surrounded by the upper, lower, left, and right boundary positions in a predetermined order. it can.
A small character is finer than a large character, and when recognition is performed by giving the same recognition point to all characters, the number of recognition points is large and processing takes time, but in the embodiment of the present invention, Since the pixel data level of a predetermined location set in advance in the cutout range is determined in order, the corresponding characters are determined and determined, and it becomes possible to omit recognition of unnecessary determination points. Processing with high recognition accuracy is possible. The coordinates of the discrimination points in the present embodiment are a preferred example in the case of the illustrated character type and character outline, and in the case of different character types and character outlines, they are set according to the character type, outline, and the like. In the embodiment of the present invention, coordinates corresponding to a plurality of kinds of staffs can be stored in a memory in advance, and can be read out from the memory and used according to the staff used.
[0078]
【The invention's effect】
As is clear from the above description, the present invention can automatically recognize the characters and numbers of the scale collimated with the telescope, so that no artificial error occurs and more accurate surveying is possible in a shorter time. .
[Brief description of the drawings]
FIG. 1 is a block diagram showing an outline of an embodiment of a surveying instrument to which a method of the present invention is applied.
FIG. 2 is a diagram showing an example of a relationship between a collimation field of view of the surveying instrument and a scale.
FIG. 3 is a flowchart showing an outline of a process for automatically reading characters on a staff of the surveying instrument.
FIG. 4 is a diagram showing an example of an image of a staff taken by the surveying instrument.
FIG. 5 is an enlarged view of a pitch scale of a staff taken by the surveying instrument.
FIG. 6 is a diagram showing in more detail the scale on the same pitch scale.
FIG. 7 is a diagram showing a state when a pattern is recognized by the surveying instrument.
FIG. 8 is a diagram showing a large character cutout range and a discrimination point when a large character is recognized by the surveying instrument.
FIG. 9 is a diagram schematically illustrating an algorithm for recognizing numbers other than “1” in the present embodiment.
FIG. 10 is a flowchart showing processing based on the algorithm schematically shown.
[Explanation of symbols]
10 Digital level
11 collimating telescope
13 Focus plate
15 Lens drive mechanism
17 Motor
19 Encoder
21 Area sensor
23 S / H / A / D conversion circuit
25 Image processing circuit
27 Frame memory (RAM)
29 Timing Generator
31 CPU (calculation means, analysis means)
33 Display
35 Memory (memory means)
37 Motor driver
51 fields of view
52 Vertical Line
53 horizontal lines
54 Upper Stadium Line
55 Roastasia Line
101 Standard
102 Long distance pattern
103 large string
104 Small string
105 Pitch scale (scale)

Claims (8)

Imaging means for collimating with a collimating telescope and imaging a scale surface of a scale with characters and black circles of a predetermined pattern positioned above the character marks at predetermined intervals ;
Storage means for storing an image of the scale surface imaged by the imaging means as pixel data in pixel units;
Analyzing means for analyzing the pixel data stored in the storage means ,
The analyzing means detects the upper end position of the character from the pixel data stored in the storage means, detects the level of the pixel data string in a horizontal direction from a predetermined position above the upper end position, and continuously black It is determined whether or not a level exists, and when one or more continuous black levels exist, the level of the pixel data string is detected vertically upward from the approximate center position of each of the continuous black levels. When there are a plurality of pixels, the level of the pixel data string is detected vertically upward from a substantially intermediate position between the continuous black level and the continuous black level, and the continuous black level detected by detecting the level of the pixel data string is detected. Detecting the number of bullets by number,
Surveying instrument equipped with an automatic character reading device for a staff characterized by The scale surface of the scale has a pitch scale in which scales made of light and dark bands having vertical and horizontal sides are arranged in a predetermined cycle in the vertical direction, and along the pitch scale on one side across the pitch scale. The character and the black circle pattern attached at a predetermined interval, and the character smaller than the character arranged on the other side at a different interval from the character and black circle pattern, is to determine the number of pixels the image width or image height of the scale of the image of the pitch scale recognizes the characters and black circle pattern or to recognize the small characters than whether crab or greater than a predetermined threshold value A surveying instrument provided with the automatic character reading device for a staff according to claim 1. When recognizing the character and the black circle pattern, the analyzing means detects the level of the pixel data in order for the pixel data sequence in the horizontal direction in the pixel data stored in the storage means, and the left and right boundary positions where the character is included Is detected for pixel data strings at different positions in the vertical direction, and when a pixel data string including left and right boundaries that contain characters is detected continuously for the height of the character, the boundary is detected. A surveying instrument provided with the automatic character reading device for a scale according to claim 1 or 2, wherein the pixel data strings at the uppermost position and the lowermost position in the vertical direction are the upper end position and the lower end position of the character. The analysis means detects whether or not there is a continuous black level in a row of pixel data for the horizontal and vertical pixel data sequences at the detected upper, lower, left and right boundary positions. When it is detected that a black level exists, the detection range of the pixel data string to be detected is expanded , and when it is determined that a continuous black level is not included, the detection range of the pixel data string is narrowed so that the character to be recognized is vertically and horizontally A surveying instrument provided with the automatic character reading device for a staff according to claim 3 for correcting the boundary position of. Said analysis means, said detected seeking vertically and horizontally aspect ratio of the character from the boundary position of the surveying instrument equipped with character automatic reader Leveling rod according to claim 4, wherein the character is determined whether 1. When the analyzing means recognizes that the character is not the number 1, it detects the pixel data level of the characteristic position corresponding to the four vertices of a substantially rectangular shape located inward from the boundary position, surveying instrument equipped with character automatic reader Leveling rod according to claim 5, wherein determining the other numbers combination thereof. When the analyzing unit determines that the character to be determined is a small one to three digit number, the analysis unit sets an approximate range of the character of each digit on the basis of a predetermined position of the scale of the pitch scale, Check the pixel data levels on the top, bottom, left and right sides surrounding the approximate range, detect the top / bottom and left / right border positions based on whether or not there are continuous black levels, and determine the character's aspect ratio based on the border positions. look, surveying instrument the character with a character automatic reader Leveling rod according to claim 2, wherein determining one or not numbers. Said analyzing means, when the character is determined not to be the number 1 calculates one feature position located inward from the boundary position, by checking the pixel data level of the feature position, the the numbers of possible as a result to the decision result, such a process of checking Similarly pixel data level to calculate the different feature positions the determination result to the original, one of numbers of possible until A surveying instrument equipped with the automatic character reading device for a staff according to claim 7 which is repeated .

Images