JPH11203450A - Image compositing device - Google Patents

Image compositing device

Info

Publication number
JPH11203450A
JPH11203450A JP10007298A JP729898A JPH11203450A JP H11203450 A JPH11203450 A JP H11203450A JP 10007298 A JP10007298 A JP 10007298A JP 729898 A JP729898 A JP 729898A JP H11203450 A JPH11203450 A JP H11203450A
Authority
JP
Japan
Prior art keywords
label
image
character recognition
images
label feature
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.)
Granted
Application number
JP10007298A
Other languages
Japanese (ja)
Other versions
JP3316445B2 (en
Inventor
Kyoko Wada
恭子 和田
Koji Suga
弘二 菅
Shintaro Kumano
信太郎 熊野
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Mitsubishi Heavy Industries Ltd
Original Assignee
Mitsubishi Heavy Industries Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Mitsubishi Heavy Industries Ltd filed Critical Mitsubishi Heavy Industries Ltd
Priority to JP00729898A priority Critical patent/JP3316445B2/en
Publication of JPH11203450A publication Critical patent/JPH11203450A/en
Application granted granted Critical
Publication of JP3316445B2 publication Critical patent/JP3316445B2/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Abstract

PROBLEM TO BE SOLVED: To prevent the generation of a failure in image composition or discontinuous surface which may cause misrecognition in an image processor for picking up an image of a three-dimensional(3D) object e.g. and measuring character information (e.g. a container number) written on the object. SOLUTION: The image compositing device is provided with a maximum correlation point retrieving device 1 for searting a maximum correlation point to be a shifted position having high correlation when two images are superposed to each other, two label feature value calculation means 2 for independently processing respective images and calculating respective label feature values, two character recognition devices 3 for applying character recognition to the label feature values found out by the devices 2, a lattice coordinate preparing device 4 for preparing virtual lattice coordinates from the maximum correlation point found by the maximum correlation point retrieving device 1, the label feature values found out by the devices 2 and symbol information to be character recognition results found out by the devices 3, an optimum superposed position searching device 5 for searching an optimum superposed position from the grating coordinates prepared by the device 4, and a label selector 6 for generating a composite image by selecting either one of two images in each label based on the symbol information.

Description

【発明の詳細な説明】DETAILED DESCRIPTION OF THE INVENTION

【0001】[0001]

【発明の属する技術分野】本発明は、画像合成装置に関
する。例えば、3次元物体を撮像して物体に書かれた文
字情報(例えばコンテナ番号など)を計測する画像処理
装置に適用可能である。
[0001] The present invention relates to an image synthesizing apparatus. For example, the present invention is applicable to an image processing apparatus that captures a three-dimensional object and measures character information (for example, a container number) written on the object.

【0002】[0002]

【従来の技術】認識対象が1台のカメラ視野におさまら
ず、複数台カメラの画像で計測する必要がある場合、従
来は、図2に示すように、相関最高点探索装置01によ
り、2枚の画像を重ねたときに相関の高いずらし位置
(相関最高点)を探索し、画素毎の画像合成装置02に
より、連続する二つの入力画像A,Bを画素レベルで合
成してから、画像処理(画像合成)を行うのが一般的で
ある。
2. Description of the Related Art In the case where the object to be recognized does not fit in the field of view of one camera and it is necessary to measure with images of a plurality of cameras, conventionally, as shown in FIG. After the superimposition of the images is performed, a shift position having a high correlation (highest correlation point) is searched, and two consecutive input images A and B are synthesized at the pixel level by the image synthesizing unit 02 for each pixel, and then image processing Generally, (image composition) is performed.

【0003】例えば、コンテナを側面から撮像した二つ
の入力画像に基づいて、コンテナ番号を認識する場合、
図3に従い次のような処理を行う。先ず、2枚の入力画
像A,Bの相関の高い点を相関最高点探索装置により、
探索する。
For example, when recognizing a container number based on two input images of a container taken from the side,
The following processing is performed according to FIG. First, a point having a high correlation between the two input images A and B is determined by a correlation maximum point searching device.
Explore.

【0004】次に、相関最高点を基に、2枚の入力画像
A,Bを画素毎に合成する(画素レベルでの合成)。引
き続き、2値化・フィルタリング等のラベル特徴量計算
により、コンテナ番号のラベルを抽出して文字認識処理
を行う。
Next, two input images A and B are synthesized for each pixel based on the highest correlation point (synthesis at a pixel level). Subsequently, the label of the container number is extracted by a label feature amount calculation such as binarization and filtering to perform a character recognition process.

【0005】[0005]

【発明が解決しようとする課題】従来手法は、画素レベ
ルでの合成を行うため、図4に示すように、カメラに写
る3次元物体の形状が異なり、正しく画像合成を行えな
い場合があり、また、図5に示すように、合成した画像
に不連続な面が残る、という問題点がある。
In the conventional method, since the synthesis at the pixel level is performed, as shown in FIG. 4, the shape of the three-dimensional object shown in the camera is different, and the image synthesis may not be performed correctly. Further, as shown in FIG. 5, there is a problem that a discontinuous surface remains in the synthesized image.

【0006】例えば、コンテナを側面から撮像した画像
を用いてコンテナ番号を認識する場合、コルゲート(側
面にある凹凸)上に書かれた文字を撮像するとカメラ設
置位置により画像中での大きさや明るさが異なり合成点
が求められなかったり、合成しても画像に不連続な面が
生じたりして、文字認識が行えない場合がある。
For example, in the case of recognizing a container number using an image of a container taken from the side, when a character written on a corrugate (unevenness on the side) is taken, the size and brightness in the image depend on the camera installation position. However, there is a case where the character recognition cannot be performed because the combining point is not obtained or the image has a discontinuous surface even when the combining is performed.

【0007】[0007]

【課題を解決するための手段】上記課題を解決する本発
明の画像合成装置は、カメラで撮像した複数画像に渡る
3次元物体に書かれた文字に対する計測処理において、
2枚の画像を重ねたときに相関の高いずらし位置である
相関最高点を探索する相関最高点探索装置と、それぞれ
の画像を独立に処理してラベル特徴量を計算するラベル
特徴量計算装置と、前記ラベル特徴量計算装置で求めた
ラベルに対して文字認識を行う文字認識装置と、前記相
関最高点探索装置で求めた相関最高点と、前記ラベル特
徴量計算装置及び文字認識装置で求めたラベル特徴量及
び文字認識結果であるシンボル情報から仮想的な格子座
標を作成する格子座標作成装置と、前記格子座標作成装
置で作成した格子座標から最適な重ね合わせ位置を探索
する最適重ね合わせ位置探索装置と、前記シンボル情報
に基づき、複数画像の一方をラベル毎に選択すること
で、合成画像を生成するラベル選択装置とを具備するこ
とを特徴とする。
An image synthesizing apparatus according to the present invention for solving the above-mentioned problems is provided in a measurement process for characters written on a three-dimensional object over a plurality of images taken by a camera.
A correlation maximum point search device that searches for a correlation maximum point that is a shifted position having a high correlation when two images are superimposed, and a label feature amount calculation device that calculates a label feature amount by independently processing each image. A character recognition device that performs character recognition on the label obtained by the label feature calculation device, a correlation maximum point obtained by the correlation maximum search device, and a character recognition device obtained by the label feature calculation device and the character recognition device. A grid coordinate creating device for creating virtual grid coordinates from label information and symbol information as a character recognition result, and an optimal overlay position search for searching for an optimal overlay position from the grid coordinates created by the grid coordinate creating device And a label selection device that generates a composite image by selecting one of a plurality of images for each label based on the symbol information.

【0008】〔作用〕本発明では、カメラで撮像した複
数画像に渡る、3次元物体に書かれた文字に対する計測
処理において、ラベル特徴量計算装置、文字認識装置で
ラベル特徴量及び文字認識結果を求め、これらシンボル
情報から格子座標作成装置により仮想的な格子座標を作
成し、最適重ね合わせ位置探索装置により、仮想的な格
子座標を用いて重ね合わせ位置を決めるので、画素レベ
ルで正しく合成できない場合にも正しく画像合成が可能
である。
According to the present invention, in a measurement process for a character written on a three-dimensional object over a plurality of images captured by a camera, the label feature amount calculation device and the character recognition device use the label feature amount and the character recognition result. Virtual grid coordinates are created by the grid coordinate creation device from these symbol information, and the overlay position is determined using the virtual grid coordinates by the optimal overlay position search device. It is also possible to correctly combine images.

【0009】また、ラベル選択装置により、シンボル情
報に基づき、複数画像の一方をラベル毎に選択すること
で合成画像を生成することにより、合成した画面に不連
続な面が残らない。このように、本発明では、画像毎の
シンボル情報に基づいて画像間の対応を図ることによ
り、3次元形状のもつ画像情報の変化の影響を受けにく
い画像の合成が可能となる。例えば、コンテナ番号認識
の場合には、ラベル特徴量と文字認識結果を利用してラ
ベル同士の相関計算を行い画像間の対応を図ることによ
り、認識を行うことが可能となる。
In addition, the label selection device generates a composite image by selecting one of a plurality of images for each label based on the symbol information, so that a discontinuous surface does not remain on the composite screen. As described above, according to the present invention, by associating the images based on the symbol information for each image, it is possible to synthesize images that are not easily affected by changes in image information having a three-dimensional shape. For example, in the case of container number recognition, recognition can be performed by calculating correlation between labels using the label feature amount and the result of character recognition to achieve correspondence between images.

【0010】[0010]

【発明の実施の形態】以下、本発明の実施の形態につい
て、図面に示す実施例を参照して詳細に説明する。本発
明の一実施例に係る画像合成装置を図1に示す。この画
像合成装置は、カメラで撮像した複数画像に渡る、3次
元物体に書かれた文字に対する計測処理、特に、コンテ
ナ番号認識という具体的課題を解決するものである。
Embodiments of the present invention will be described below in detail with reference to embodiments shown in the drawings. FIG. 1 shows an image synthesizing apparatus according to an embodiment of the present invention. This image synthesizing apparatus solves a specific problem of measurement processing for characters written on a three-dimensional object over a plurality of images captured by a camera, particularly, container number recognition.

【0011】即ち、この画像処理装置は、2枚の画像を
重ねたときに相関の高いずらし位置である相関最高点を
探索する相関最高点探索装置1と、それぞれの画像を独
立に処理してラベル特徴量を計算するラベル特徴量計算
装置2,2と、前記ラベル特徴量計算装置2,2で求め
たラベルに対して文字認識を行う文字認識装置3,3
と、前記相関最高点探索装置1で求めた相関最高点と、
前記ラベル特徴量計算装置2,2及び文字認識装置3,
3で求めたラベル特徴量及び文字認識結果であるシンボ
ル情報から仮想的な格子座標を作成する格子座標作成装
置4と、前記格子座標作成装置4で作成した格子座標か
ら最適な重ね合わせ位置を探索する最適重ね合わせ位置
探索装置5と、前記シンボル情報に基づき、複数画像の
一方をラベル毎に選択することで、合成画像を生成する
ラベル選択装置6とを具備する。
That is, this image processing apparatus independently searches the highest correlation point searching apparatus 1 for searching for the highest correlation point which is a shift position having a high correlation when two images are superimposed, and processes each image independently. Label feature calculation devices 2 and 2 for calculating label feature, and character recognition devices 3 and 3 for performing character recognition on the labels obtained by the label feature calculation devices 2 and 2
And the highest correlation point obtained by the highest correlation point search device 1;
The label feature amount calculation devices 2 and 2 and the character recognition device 3
3. A grid coordinate creating device 4 for creating virtual grid coordinates from the label feature quantity obtained in step 3 and the symbol information as the character recognition result, and searching for an optimal superposition position from the grid coordinates created by the grid coordinate creating device 4. And a label selection device 6 for generating a composite image by selecting one of a plurality of images for each label based on the symbol information.

【0012】相関最高点探索装置1は、2つの入力画像
A,Bのある画像領域を重ねたときに相関の高いずらし
位置を2枚の入力画像A,Bの相関最高点とする。ラベ
ル特徴量計算装置2は、それぞれの入力画像A,Bのフ
ィルタリング、2値化を行い、ラベルを抽出してラベル
特徴量計算を行う。
The correlation maximum point searching device 1 sets a shift position having a high correlation when a certain image area of the two input images A and B is overlapped, as a correlation maximum point of the two input images A and B. The label feature amount calculation device 2 performs filtering and binarization of each of the input images A and B, extracts a label, and performs label feature amount calculation.

【0013】文字認識装置3は、ラベル特徴量計算装置
2で求めたラベルに対して文字認識処理を行う。
The character recognition device 3 performs a character recognition process on the label obtained by the label feature amount calculation device 2.

【0014】格子座標作成装置4は、相関最高点探索装
置1で求めた相関最高点と、ラベル特徴量計算装置2で
求めたラベル特徴量及び文字認識装置3で求めた文字認
識結果であるシンボル情報から仮想的な格子座標を作成
する。
The grid coordinate creation device 4 includes a correlation maximum point obtained by the correlation maximum point search device 1, a label feature amount obtained by the label feature amount calculation device 2, and a symbol which is a character recognition result obtained by the character recognition device 3. Create virtual grid coordinates from the information.

【0015】具体的には、図6に示すように、左画像と
右画像とについて相関最高点探索を行い、相関最高点
(x,y)を基準にして、y方向については、yラベル
中心の頻度分布領域の対応をとり、x方向については、
ラベル数毎に区切ることにより、図中破線で示す格子座
標を作成する。
More specifically, as shown in FIG. 6, the highest correlation point is searched for the left image and the right image, and the center of the y label is determined in the y direction based on the highest correlation point (x, y). Of the frequency distribution area, and in the x direction,
A grid coordinate indicated by a broken line in the figure is created by dividing each label number.

【0016】最適重ね合わせ位置探索装置5は、ラベル
特徴量計算装置2、文字認識装置3で求めたシンボル情
報及び格子座標作成装置4で作成した格子座標から、画
像同士の最適な重ね合わせの位置を探索する。これは、
(1)式を一例として、その最高点を持つ座標を計算し
て最適重ね合わせ位置を求める。(1)式は、文字種の
一致数の多さ、不一致数の少なさを評価する関数の例で
ある。
The optimum superposition position searching device 5 calculates the optimum superposition position between the images based on the symbol information obtained by the label feature amount calculation device 2 and the character recognition device 3 and the grid coordinates generated by the grid coordinate generation device 4. To explore. this is,
Using the formula (1) as an example, the coordinates having the highest point are calculated to obtain the optimum superimposition position. Equation (1) is an example of a function that evaluates the number of matches of character types and the number of mismatches.

【0017】 評価値E=C1+α・C2+β・C3 …(1) 但し、C1:右画像のラベルの認識1位と左又は下画像
のラベルの認識1〜3位が一致する文字数 C2:|Dist1−Dist2|<Th1となる文字数 C3:|Dist1−Dist2|>Th2となる文字数 α,β:重み係数 Dist1:右画像のラベルの認識1位の文字認識得点 Dist2:左又は下画像のラベルの認識1位の文字認識得
点 Th1,Th2:閾値
Evaluation value E = C 1 + α · C 2 + β · C 3 (1) where C 1 : the first recognition of the label of the right image matches the first to third recognition of the label of the left or lower image. characters C 2: | D ist1 -D ist2 | <T h1 become characters C 3: | D ist1 -D ist2 |> T h2 become characters alpha, beta: weight coefficient D IST1: recognition 1-position of the right image label character recognition score D Ist2: character recognition score of the recognition position 1 of label left or lower image T h1, T h2: threshold

【0018】具体的には、図7のフローチャートに従
い、以下のように、ラベル特徴(=シンボル)を用いて
相互相関の計算を行い、2枚の画像のマッチングを行
う。先ず、各入力画像の格子座標、各入力画像の文字認
識結果に基づき、格子座標の文字認識結果を1段ずつ取
り出して1文字分ずつ重ね合わせる(ステップS1)。
More specifically, according to the flowchart of FIG. 7, a cross-correlation is calculated using a label feature (= symbol) as follows, and matching between two images is performed. First, based on the grid coordinates of each input image and the character recognition results of each input image, the character recognition results of the grid coordinates are extracted one by one and superimposed one character at a time (step S1).

【0019】格子座標の段は、図9に示すように、上か
ら下へ、1段、2段、3段…と定義し、ラベルは、左右
方向へ、1、2、3…と定義する。例えば、横書きの場
合、図9に示すように、左画像の格子座標及び文字認識
結果と右画像の格子座標及び文字認識結果は、格子座標
を境にして上下左右に並ぶので、1段ずつ取り出して、
先ず、1文字分重ね合わせ、次に、左右にずらして2文
字分重ね合わせ、更に、3文字分以降も引き続き同様に
重ね合わせを行う。
As shown in FIG. 9, the steps of the grid coordinates are defined as 1, 2, 3,... From top to bottom, and the labels are defined as 1, 2, 3,. . For example, in the case of horizontal writing, as shown in FIG. 9, the grid coordinates and character recognition results of the left image and the grid coordinates and character recognition results of the right image are lined up, down, left, and right with the grid coordinates as boundaries. hand,
First, one character is superimposed, then two characters are superimposed to be shifted to the left and right, and the same is repeated for three characters and thereafter.

【0020】次に、重ね合わせの評価を行う(ステップ
S2)。上記例では、図10に示すように、各段を2文
字分重ね合わせたときに、重ね合わせの評価値が最高と
なる。更に同様な手順を全ての段について繰り返す。
尚、段数のループはi=0から始めるものとし、左右の
ラベル数のループはj=0(j:1…m),k=n
(k:1…n)から始めるものとする。引き続き、重ね
合わせの評価値が最高のものを重ね合わせ位置とする
(ステップS3)。
Next, the superposition is evaluated (step S2). In the above example, as shown in FIG. 10, when each row is overlapped by two characters, the evaluation value of the overlap is the highest. Further, the same procedure is repeated for all stages.
The loop of the number of stages starts from i = 0, the loop of the number of right and left labels is j = 0 (j: 1... M), and k = n.
(K: 1... N). Subsequently, the one with the highest evaluation value of the overlay is set as the overlay position (step S3).

【0021】ラベル選択装置6は、最適な重ね合わせ位
置においてシンボル情報に基づき複数画像の一方をラベ
ル毎に選択するラベルの選択を行う。具体的には、図8
のフローチャートに従い、以下のように、ラベルのシン
ボル情報に基づき、被数画像の一方をラベル毎に選択す
ることで合成画像を生成する。
The label selection device 6 selects a label for selecting one of a plurality of images for each label based on the symbol information at the optimum overlapping position. Specifically, FIG.
According to the flowchart of FIG. 7, a composite image is generated by selecting one of the algebraic images for each label based on the symbol information of the label as follows.

【0022】先ず、最適重ね合わせ位置、各入力画像の
格子座標及び各入力画像のラベル特微量・文字認織結果
に基づき、2枚の画像を重ね合わせる(ステップT
1)。次に、ラベルが重ならないときには、ラベル特徴
量・文字認証結果を1つの座標系にコピーする(ステッ
プT2)。また、ラベルが重なるときには、文字認識得
点の高いラベルを選択する(ステップT3)。このよう
な手順を全ての段について繰り返すことにより、文字列
を抽出する。尚、段数のループは、i=0から始める。
First, two images are superimposed on the basis of the optimum superimposition position, the grid coordinates of each input image, and the results of labeling and character recognition of each input image (step T).
1). Next, when the labels do not overlap, the label feature amount / character authentication result is copied to one coordinate system (step T2). When labels overlap, a label having a high character recognition score is selected (step T3). By repeating such a procedure for all stages, a character string is extracted. Note that the loop of the number of stages starts from i = 0.

【0023】例えば、横書きの場合、図10に示すよう
に、格子座標を基に重ね合わせ、ラベルが重なるときに
は、文字認識得点の高い方のラベル画像を選択してコピ
ーする。
For example, in the case of horizontal writing, as shown in FIG. 10, when the labels are overlapped based on the lattice coordinates, and the labels overlap, the label image with the higher character recognition score is selected and copied.

【0024】上記構成を有する本実施例に係る画像合成
装置では、カメラで撮像した複数画像に渡る、3次元物
体に書かれた文字に対する計測処理において、ラベル特
徴量計算装置2、文字認識装置3でラベル特徴量及び文
字認識結果を求め、これらシンボル情報から格子座標作
成装置4により仮想的な格子座標を作成し、最適重ね合
わせ位置探索装置5により、仮想的な格子座標を用いて
重ね合わせ位置を決めるので、画素レベルで正しく合成
できない場合にも正しく画像合成が可能である。
In the image synthesizing apparatus according to the present embodiment having the above configuration, in the measurement processing for characters written on a three-dimensional object over a plurality of images captured by a camera, the label feature amount calculating device 2 and the character recognizing device 3 , A label feature amount and a character recognition result are obtained, a virtual grid coordinate is created by the grid coordinate creation device 4 from the symbol information, and an overlay position is searched for by the optimal overlay position search device 5 using the virtual grid coordinates. Is determined, the image can be correctly synthesized even when the image cannot be correctly synthesized at the pixel level.

【0025】また、ラベル選択装置6により、シンボル
情報に基づき、複数画像の一方をラベル毎に選択するこ
とで合成画像を生成することにより、合成した画面に不
連続な面が残らない。従って、本実施例では、画像毎の
シンボル情報に基づいて画像間の対応を図ることによ
り、3次元形状のもつ画像情報の変化の影響を受けにく
い画像の合成が可能となる。例えば、コンテナ番号認識
の場合には、ラベル特徴量と文字認識結果を利用してラ
ベル同士の相関計算を行い画像間の対応を図ることによ
り、認識を行うことが可能となる。
Further, the label selection device 6 generates a composite image by selecting one of a plurality of images for each label based on the symbol information, so that a discontinuous surface does not remain on the composite screen. Therefore, in the present embodiment, by associating the images based on the symbol information for each image, it is possible to synthesize images that are not easily affected by changes in image information having a three-dimensional shape. For example, in the case of container number recognition, recognition can be performed by calculating correlation between labels using the label feature amount and the result of character recognition to achieve correspondence between images.

【0026】[0026]

【発明の効果】以上、実施例に基づいて具体的に説明し
たように、本発明の請求項1に係る画像合成装置は、カ
メラで撮像した複数画像に渡る3次元物体に書かれた文
字に対する計測処理において、2枚の画像を重ねたとき
に相関の高いずらし位置である相関最高点を探索する相
関最高点探索装置と、それぞれの画像を独立に処理して
ラベル特徴量を計算するラベル特徴量計算装置と、前記
ラベル特徴量計算装置で求めたラベルに対して文字認識
を行う文字認識装置と、前記相関最高点探索装置で求め
た相関最高点と、前記ラベル特徴量計算装置及び文字認
識装置で求めたラベル特徴量及び文字認識結果であるシ
ンボル情報から仮想的な格子座標を作成する格子座標作
成装置と、前記格子座標作成装置で作成した格子座標か
ら最適な重ね合わせ位置を探索する最適重ね合わせ位置
探索装置と、前記シンボル情報に基づき、複数画像の一
方をラベル毎に選択することで、合成画像を生成するラ
ベル選択装置とを具備するので、画像で認識対象を発見
する際に、認識間違いの原因となる画像合成の失敗や、
不連続面が生じるのを防ぐことができる。
As described above in detail with reference to the embodiments, the image synthesizing apparatus according to the first aspect of the present invention is capable of processing characters written on a three-dimensional object over a plurality of images captured by a camera. In the measurement processing, a correlation maximum point search device that searches for a correlation maximum point that is a shift position having a high correlation when two images are superimposed, and a label feature that independently processes each image to calculate a label feature amount An amount calculating device, a character recognition device that performs character recognition on the label obtained by the label feature amount calculating device, a correlation maximum point obtained by the correlation maximum point searching device, the label feature amount calculating device, and the character recognition. A grid coordinate creating device for creating virtual grid coordinates from the label information obtained by the device and the symbol information as the character recognition result, and an optimal superimposition from the grid coordinates created by the grid coordinate creating device. Since there is provided an optimal superposition position searching device for searching for a position, and a label selecting device for generating a composite image by selecting one of a plurality of images for each label based on the symbol information, the recognition target is determined by the image. When discovering, image synthesis failures that cause recognition errors,
The generation of a discontinuous surface can be prevented.

【図面の簡単な説明】[Brief description of the drawings]

【図1】本発明一実施例に係る画像合成装置を示す全体
構成図である。
FIG. 1 is an overall configuration diagram showing an image synthesizing apparatus according to an embodiment of the present invention.

【図2】従来技術の概略構成図である。FIG. 2 is a schematic configuration diagram of a conventional technique.

【図3】従来技術に係る画像合成手法の説明図である。FIG. 3 is an explanatory diagram of an image synthesis method according to the related art.

【図4】従来技術に係る問題点の説明図である。FIG. 4 is an explanatory diagram of a problem according to the related art.

【図5】従来技術に係る問題点の説明図である。FIG. 5 is an explanatory diagram of a problem according to the related art.

【図6】仮想的な格子座標の作成方法を示すフローチャ
ートである。
FIG. 6 is a flowchart illustrating a method of creating virtual grid coordinates.

【図7】最適な重ね合わせの位置の探索方法を示すフロ
ーチャートである。
FIG. 7 is a flowchart illustrating a method of searching for an optimum position of superposition.

【図8】ラベル選択の方法を示す説明図である。FIG. 8 is an explanatory diagram showing a label selection method.

【図9】格子座標の段とラベルの定義及び重ね合わせの
評価方法を示す説明図である。
FIG. 9 is an explanatory diagram showing a method of evaluating the definition and superimposition of a grid coordinate stage and a label.

【図10】2枚の画像の重ね合わせを示す説明図であ
る。
FIG. 10 is an explanatory diagram showing superposition of two images.

【符号の説明】[Explanation of symbols]

1 相関最高点探索装置 2 ラベル特徴量計算装置 3 文字認識装置 4 格子座標作成装置 5 位置探索装置 6 ラベル選択装置 A,B 入力画像 DESCRIPTION OF SYMBOLS 1 Correlation highest point search device 2 Label feature amount calculation device 3 Character recognition device 4 Grid coordinate creation device 5 Position search device 6 Label selection device A, B Input image

Claims (1)

【特許請求の範囲】[Claims] 【請求項1】 カメラで撮像した複数画像に渡る3次元
物体に書かれた文字に対する計測処理において、2枚の
画像を重ねたときに相関の高いずらし位置である相関最
高点を探索する相関最高点探索装置と、それぞれの画像
を独立に処理してラベル特徴量を計算するラベル特徴量
計算装置と、前記ラベル特徴量計算装置で求めたラベル
に対して文字認識を行う文字認識装置と、前記相関最高
点探索装置で求めた相関最高点と、前記ラベル特徴量計
算装置及び文字認識装置で求めたラベル特徴量及び文字
認識結果であるシンボル情報から仮想的な格子座標を作
成する格子座標作成装置と、前記格子座標作成装置で作
成した格子座標から最適な重ね合わせ位置を探索する最
適重ね合わせ位置探索装置と、前記シンボル情報に基づ
き、複数画像の一方をラベル毎に選択することで、合成
画像を生成するラベル選択装置とを具備することを特徴
とする画像合成装置。
1. In a measurement process for characters written on a three-dimensional object over a plurality of images captured by a camera, a correlation maximum searching for a correlation maximum point that is a shift position having a high correlation when two images are superimposed. A point search device, a label feature value calculation device that independently processes each image to calculate a label feature value, a character recognition device that performs character recognition on the label obtained by the label feature value calculation device, A grid coordinate creating apparatus that creates virtual grid coordinates from the highest correlation point obtained by the highest correlation point searching device, the label feature amounts obtained by the label feature amount calculating device and the character recognition device, and the symbol information as the character recognition result. An optimal superposition position searching device that searches for an optimal superposition position from the lattice coordinates created by the lattice coordinate creating device; and one of a plurality of images based on the symbol information. And a label selection device for generating a composite image by selecting a label for each label.
JP00729898A 1998-01-19 1998-01-19 Image synthesis device Expired - Fee Related JP3316445B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP00729898A JP3316445B2 (en) 1998-01-19 1998-01-19 Image synthesis device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP00729898A JP3316445B2 (en) 1998-01-19 1998-01-19 Image synthesis device

Publications (2)

Publication Number Publication Date
JPH11203450A true JPH11203450A (en) 1999-07-30
JP3316445B2 JP3316445B2 (en) 2002-08-19

Family

ID=11662129

Family Applications (1)

Application Number Title Priority Date Filing Date
JP00729898A Expired - Fee Related JP3316445B2 (en) 1998-01-19 1998-01-19 Image synthesis device

Country Status (1)

Country Link
JP (1) JP3316445B2 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009077537A1 (en) * 2007-12-19 2009-06-25 Societe De Technologie Michelin Method of evaluation by comparison of an acquired image with a reference image
WO2011065065A1 (en) * 2009-11-25 2011-06-03 シャープ株式会社 Information processing device and method for controlling an information processing device
CN105592792A (en) * 2013-10-23 2016-05-18 日立麦克赛尔株式会社 Display apparatus configured for display of lower resolution composite color subfields
JP2018182504A (en) * 2017-04-11 2018-11-15 富士ゼロックス株式会社 Image processing apparatus and program

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009077537A1 (en) * 2007-12-19 2009-06-25 Societe De Technologie Michelin Method of evaluation by comparison of an acquired image with a reference image
FR2925687A1 (en) * 2007-12-19 2009-06-26 Michelin Soc Tech METHOD OF EVALUATION BY COMPARISON OF AN ACQUIRED IMAGE WITH A REFERENCE IMAGE.
JP2011506989A (en) * 2007-12-19 2011-03-03 ソシエテ ド テクノロジー ミシュラン Evaluation method of collected image by comparison with reference image
US8498466B2 (en) 2007-12-19 2013-07-30 Compagnie Generale Des Etablissements Michelin Method of evaluation by comparison of an acquired image with a reference image
WO2011065065A1 (en) * 2009-11-25 2011-06-03 シャープ株式会社 Information processing device and method for controlling an information processing device
JP2011113214A (en) * 2009-11-25 2011-06-09 Sharp Corp Information processing device, and method for controlling information processing device
CN102667813A (en) * 2009-11-25 2012-09-12 夏普株式会社 Information processing device and method for controlling an information processing device
US8867840B2 (en) 2009-11-25 2014-10-21 Sharp Kabushiki Kaisha Information processing device and method for controlling an information processing device
CN105592792A (en) * 2013-10-23 2016-05-18 日立麦克赛尔株式会社 Display apparatus configured for display of lower resolution composite color subfields
CN105592792B (en) * 2013-10-23 2018-08-17 麦克赛尔控股株式会社 Surface state measures analytical information management system and surface state measures analysis approaches to IM
JP2018182504A (en) * 2017-04-11 2018-11-15 富士ゼロックス株式会社 Image processing apparatus and program

Also Published As

Publication number Publication date
JP3316445B2 (en) 2002-08-19

Similar Documents

Publication Publication Date Title
US7256799B2 (en) Image synthesizer, image synthesis method and computer readable recording medium having image synthesis processing program recorded thereon
CN107990899A (en) A kind of localization method and system based on SLAM
CN110163903A (en) The acquisition of 3-D image and image position method, device, equipment and storage medium
CN110517293A (en) Method for tracking target, device, system and computer readable storage medium
JP6372149B2 (en) Display control apparatus, display control method, and display control program
KR20090108642A (en) Method of capturing, processing, and rendering images
JP3549569B2 (en) Target pattern detection method in video
KR20160044999A (en) System and Method for Searching Choreography Database based on Motion Inquiry
CN109493278A (en) A kind of large scene image mosaic system based on SIFT feature
CN108648194A (en) Based on the segmentation of CAD model Three-dimensional target recognition and pose measuring method and device
JP3571162B2 (en) Similar object search method and apparatus
JP3411469B2 (en) Frame multiplex image creation method
CN110544202A (en) parallax image splicing method and system based on template matching and feature clustering
JP2005339127A (en) Apparatus and method for displaying image information
JP2003061936A (en) Moving three-dimensional model formation apparatus and method
CN107133260A (en) The matching and recognition method and device of a kind of landmark image
JP3316445B2 (en) Image synthesis device
JPH10301948A (en) Method for retrieving image and video
JP2004198530A (en) Map updating system, map updating method and computer program
CN104104911B (en) Timestamp in panoramic picture generating process is eliminated and remapping method and system
CN107436943A (en) A kind of live shoe sole print search method of combination subjectivity conspicuousness
CN106228553A (en) High-resolution remote sensing image shadow Detection apparatus and method
CN107273937A (en) Target any angle labeling method in a kind of video and image
CN107784688A (en) A kind of three-dimensional modeling method based on picture
CN107067009A (en) A kind of real-time bar recognition methods

Legal Events

Date Code Title Description
A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

Effective date: 20020514

LAPS Cancellation because of no payment of annual fees