JP4578638B2 - Image recognition method - Google Patents

Image recognition method Download PDF

Info

Publication number
JP4578638B2
JP4578638B2 JP2000233816A JP2000233816A JP4578638B2 JP 4578638 B2 JP4578638 B2 JP 4578638B2 JP 2000233816 A JP2000233816 A JP 2000233816A JP 2000233816 A JP2000233816 A JP 2000233816A JP 4578638 B2 JP4578638 B2 JP 4578638B2
Authority
JP
Japan
Prior art keywords
image
closed curve
recognition
energy
recognition object
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
JP2000233816A
Other languages
Japanese (ja)
Other versions
JP2002049922A (en
Inventor
昇 東
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.)
Panasonic Corp
Panasonic Holdings Corp
Original Assignee
Panasonic Corp
Matsushita Electric Industrial Co 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 Panasonic Corp, Matsushita Electric Industrial Co Ltd filed Critical Panasonic Corp
Priority to JP2000233816A priority Critical patent/JP4578638B2/en
Publication of JP2002049922A publication Critical patent/JP2002049922A/en
Application granted granted Critical
Publication of JP4578638B2 publication Critical patent/JP4578638B2/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Landscapes

  • Image Analysis (AREA)

Description

【0001】
【発明の属する技術分野】
本発明は、画像認識方法に関するものであり、特に、動的輪郭抽出方法におけるエネルギー制御方法に関するものである。
【0002】
【従来の技術】
画像に含まれる特定の認識対象物の輪郭を抽出する方法として、動的輪郭抽出方法(Snakes: “active contour model” ,International Journal of Computer Vision ,Vol.1, No.4, pp.321-331, 1988)が知られている。
【0003】
この方法は、画像に含まれる特定の認識対象物に対して変形可能な閉曲線を設定し、その閉曲線のエネルギーを定義することにより、その閉曲線の状態を定量的に評価するものである。閉曲線のエネルギーは閉曲線が認識対象物の輪郭に一致した場合に最も小さくなるように定義されるものであり、このことから閉曲線をそのエネルギーが最小となるように変形させることで認識対象物の輪郭を抽出することができる。なお、閉曲線はその閉曲線を形成する離散的に配置された制御点を連結したものである。従って、上述の閉曲線を変形させる処理とは、閉曲線のエネルギーが最小となるように閉曲線上の各制御点を移動させることを意味する。具体的には、閉曲線の状態に対応して定義される、内部エネルギー(内部スプラインエネルギー)、画像エネルギー、および外部エネルギー、の3つのエネルギーの和が最小になるように閉曲線上の制御点v(s)を移動させることにより、認識対象物の輪郭を抽出する。この時の閉曲線のエネルギーは[数1]に示すような式で表せられる。
【数1】

Figure 0004578638
【0004】
なお、内部スプラインエネルギーは、制御点の1次微分および2次微分からなり、閉曲線を収縮させ、さらに滑らかにする力を作用させるエネルギーで、[数2]に示す式で表せられる。
【数2】
Figure 0004578638
【0005】
また、画像エネルギーは、制御点座標における画像の微分により与えられ、閉曲線を画像のエッジに張り付かせる力を作用させるエネルギーで、[数3]に示すような式で表せられる。
【数3】
Figure 0004578638
【0006】
また、外部エネルギーは、外部から閉曲線に対しエネルギーを与え、閉曲線を収縮あるいは膨張させる力を作用させるエネルギーで、[数4]に示す式で表せられる。
【数4】
Figure 0004578638
【0007】
【発明が解決しようとする課題】
従来の動的輪郭抽出方法において、閉曲線は、縮小方向への変形か、あるいは膨張方向への変形のどちらか一方向の変形しか行うことができない。よって、画像認識処理を開始する際、閉曲線を縮小方向へ変形する制御を行う場合は、認識対象物を囲むように閉曲線を配置する必要がある。また、閉曲線を膨張方向へ変形する制御を行う場合は、認識対象物の中に全ての線分が含まれるように閉曲線を初期設定する必要がある。仮に閉曲線の初期設定を認識対象領域と交差するように設定してしまうと、閉曲線は認識対象物の全周におけるエッジに貼りつかず認識に失敗する。すなわち、図3に示すように、制御点330〜343を連結して形成した初期閉曲線320を認識対象物310と交差するように設定した場合、縮小制御を行うと初期閉曲線320は閉曲線330へと変形することから、閉曲線330は認識対象物310の全周におけるエッジに貼りつかず、認識対象物310を正確に認識することができない。つまり、従来の動的輪郭抽出方法において画像中の認識対象物を認識する精度は、閉曲線の初期座標に大きく依存した結果となり、閉曲線の初期設定の状態によっては認識対象物の認識を行うことができない可能性が生じる。
【0008】
また、従来の動的輪郭抽出方法を適用したコンピュータシステムによる画像認識処理において、その処理速度を高速化する場合に生じる問題点について以下に述べる。図4に、一般的なコンピュータシステムにより画像処理を行う場合のデータフロ−を示す。コンピュータシステム400において各種画像処理を行う場合、画像処理に必要なデータは、ハードディスク440やCCDカメラ等のセンサから、メインメモリ430に読み込まれ、そのデータはCPU420に転送され、その後キャッシュメモリ410に格納される。キャッシュメモリ410に既に必要なデータが転送されている場合は、CPU420がキャッシュメモリ410からそのデータの読み込みを行う。この時、一般にキャッシュメモリ410へのアクセス速度は高速であり、これに対しメインメモリ430へのアクセス速度は低速である。しかしキャッシュメモリ410は高価であるためコンピュータシステムにおいては少量サイズしか搭載しておらず、これに対し安価なメインメモリ430は大容量搭載されているのが一般的である。
【0009】
上記コンピュータ構造から、画像処理の処理速度を高速化させるためには、高速なデータアクセスが可能なキャッシュメモリ410を有効活用することが重要となる。CPU420の必要なデータがキャッシュメモリ410に存在することを現す比率をヒット率と総称するが、このヒット率を向上させることが高速化に直結する。動的輪郭抽出方法を適用したコンピュータシステムにおいてヒット率を向上させる方法としては、画像データを小領域に分割して閉曲線を決定し、その領域毎に対象を認識することにより、処理対象データの容量を少量に抑える方法が考えられる。しかし上述のように、従来の動的輪郭抽出方法においては、閉曲線を配置する際に、その閉曲線を認識対象領域と交差しないように初期設定する必要がある。しかしながら、図5のように画像500上に認識対象物520が配置されていると、画像500を分割する分割線510は複雑な配置となり、閉曲線を容易に決定することができないという問題が生じる。また分割線510により画像を分割し、閉曲線を決定できた場合においても、この分割線の配置問題を解決するための処理時間がオーバーヘッドとなる。
【0010】
よって、本発明では、画像に含まれる特定の認識対象物の輪郭を抽出する動的輪郭抽出方法において、認識対象物に対して配置する閉曲線の初期設定条件を求めることなく、認識対象物を正確に認識することのできる画像認識方法を提供することを目的とする。
【0011】
【課題を解決するための手段】
請求項1に記載の画像認識方法は、画像上に閉曲線を配置し、曲線の滑らかさを示す内部エネルギーと、前記画像の濃度勾配を示す画像エネルギーと、外部からの外部エネルギーとを用いて、前記閉曲線の縮小処理あるいは膨張処理を行い、前記画像に含まれる認識対象物の輪郭を抽出する動的輪郭抽出方法において、前記画像は認識対象物と背景部とを含む2値画像であり、初期閉曲線を認識対象物と交差するように設定し、前記認識対象物における前記外部エネルギーを前記閉曲線が膨張する方向へ、前記背景部における前記外部エネルギーを前記閉曲線が縮小する方向へ、前記画像の濃度値に応じて動的に変化させることを特徴とする。
【0012】
請求項2に記載の画像認識方法は、画像上に閉曲線を配置し、曲線の滑らかさを示す内部エネルギーと、前記画像の濃度勾配を示す画像エネルギーと、外部からの外部エネルギーとを用いて、前記閉曲線の縮小処理あるいは膨張処理を行い、前記画像に含まれる認識対象物の輪郭を抽出する動的輪郭抽出方法において、前記画像は認識対象物と背景部とを含む2値画像であり、初期閉曲線を、複数のタイル状に分割してなる各分割領域の周縁に一致するように配置し、前記認識対象物における前記外部エネルギーを前記閉曲線が膨張する方向へ、前記背景部における前記外部エネルギーを前記閉曲線が縮小する方向へ、前記画像の濃度値に応じて動的に変化させることを特徴とする
【0014】
【発明の実施の形態】
(実施の形態1)
以下、実施の形態1に係る画像認識方法について図1を用いて説明する。
図1において、2値画像100は認識処理対象となる認識対象物110と認識対象物の外部(背景部)160からなり、認識対象物110の濃度値(画素値)を1の値に、また背景部160の画素値を0の値に設定したものである。
【0015】
以下、図1に示す2値画像100中の認識対象物110を閉曲線120により認識する方法について説明する。
まず、図1に示すように、初期閉曲線120を認識対象物110と交差するように設定する。ここで閉曲線とは、その閉曲線上に離散的に配置した制御点(130〜137,140,141)を連結したもののことを指す。この各制御点を連結する処理においては、方向性を有するものとし、半時計周りを正とする。閉曲線120の基本制御は、[数1]に示す閉曲線のエネルギーESNAKE(v)が最小となるように各制御点を移動制御することによって行う。
【0016】
次に、閉曲線120上の各制御点が、認識対象物110の内側に位置するか、または外側に位置するかの判定を行う。この判定は各制御点座標における2値画像100の画素値により行い、制御点座標における画素値が1の場合には認識対象物110の内側に位置すると判定し、一方、制御点座標における画素値が0の場合には認識対象物110の外側、すなわち背景部160に位置すると判定する。なお、図1においては、認識対象物110の外側にある制御点を黒丸(130〜137)で示し、内側にある制御点を白丸(140,141)で示している。
【0017】
次に、外部エネルギーを動的に変化させ、閉曲線120上の各制御点の移動方向を決定する。認識対象物110の外側にある制御点に対しては、対象とする制御点の前後の制御点を結ぶことで得られるベクトルを90度回転させたベクトルを外部エネルギーとして作用させる。また、認識対象物110の内側にある制御点に対しては、対象とする制御点の前後の制御点を結ぶことで得られるベクトルをマイナス90度回転させたベクトルを外部エネルギーとして作用させる。なお、ベクトルの絶対値は正規化した値を用いてもよい。
【0018】
具体的には、例えば、移動制御対象の制御点を図1に示す制御点131とした場合、制御点131は認識対象物110の外側に存在するため、制御点131の前後に位置する制御点130と制御点132とを結ぶことで得られるベクトル145を90度回転させ、外部エネルギー146を生成する。この生成した外部エネルギー146は制御点131に対して認識対象物110の外側から内側へ作用する。そして、制御点131は、外部エネルギー146に従って認識対象物110の外側から内側、すなわち閉曲線120が縮小する方向へ移動する。以上のようにして、認識対象物110の外側に位置する制御点(130〜137)に対しては認識対象物110の外側から内側に外部エネルギーを作用させ、これにより制御点(130〜137)を閉曲線120が縮小する方向に移動させるようにする。また、認識対象物110の内側に位置する制御点(140,141)に対しては、認識対象物110の内側から外側に外部エネルギーを作用させ、これにより制御点(140,141)を閉曲線120が膨張する方向へ移動させるようにする。その結果、初期閉曲線120は、閉曲線150へと変形して認識対象物310の全周におけるエッジに貼りつき、認識対象物110の輪郭が認識される。
【0019】
以上のように、実施の形態1の画像認識方法によれば、2値画像100に含まれる認識対象物110の輪郭を認識する際に、2値画像100の画像値に応じて外部エネルギーを動的に変化させ、閉曲線120上の各制御点を、この外部エネルギーに従って閉曲線120が縮小・膨張する方向に移動させるようにしたことから、閉曲線120を認識対象物110と交差するように初期設定した場合においても認識対象物110の輪郭を正確に認識することができる。
【0020】
(実施の形態2)
以下、実施の形態2に係る画像認識方法について図2を用いて説明する。
本実施の形態2に係る画像認識方法は、画像に含まれる特定の認識対象物を認識する際に、その処理速度を高速化するために、画像を小領域に分割し、領域分割毎に認識対象物を認識することを特徴とする。
【0021】
具体的には、まず、画像200を分割線210により小領域に分割し、閉曲線(231〜238)を決定する。このような処理を行うことにより、図2に示すように、画像200に対して複数の閉曲線がタイル状に配置された状態になる。
【0022】
次に、各分割領域毎に実施の形態1の画像認識方法を用いて認識対象物220の認識を行う。なお、本実施の形態2では、各分割領域毎に実施の形態1の画像認識方法を用いて認識対象物220の認識を行うことから、画像200に対して複数の閉曲線をタイル状に配置する際に、認識対象物220が閉曲線(231〜238)と交差しないように初期設定する必要がない。
【0023】
以上のように本実施の形態2の画像認識方法によれば、画像200に複数の閉曲線(231〜238)をタイル状に配置し、その領域毎に認識対象物220を実施の形態1の画像認識方法を用いて認識していることから、認識対象物220に対する閉曲線の初期設定条件を求めることなく、画像200を単純にタイル状に分割して、その領域毎に認識対象物220を認識することができ、認識対象物220の認識処理速度を向上させることが可能となる。
【0024】
【発明の効果】
以上のように請求項1に記載の画像認識方法によれば、画像上に閉曲線を配置し、曲線の滑らかさを示す内部エネルギーと、前記画像の濃度勾配を示す画像エネルギーと、外部からの外部エネルギーとを用いて、前記閉曲線の縮小処理あるいは膨張処理を行い、前記画像に含まれる認識対象物の輪郭を抽出する動的輪郭抽出方法において、前記画像は認識対象物と背景部とを含む2値画像であり、初期閉曲線を認識対象物と交差するように設定し、前記認識対象物における前記外部エネルギーを前記閉曲線が膨張する方向へ、前記背景部における前記外部エネルギーを前記閉曲線が縮小する方向へ、前記画像の濃度値に応じて動的に変化させるようにしたことから、前記閉曲線の縮小及び膨張制御を前記画像の画素値に応じて行うことができ、認識対象物に対して配置する閉曲線の初期設定条件を求めることなく、正確に画像中に含まれる認識対象物の輪郭を認識することが可能となる。
【0025】
また、請求項2に記載の画像認識方法によれば、画像上に閉曲線を配置し、曲線の滑らかさを示す内部エネルギーと、前記画像の濃度勾配を示す画像エネルギーと、外部からの外部エネルギーとを用いて、前記閉曲線の縮小処理あるいは膨張処理を行い、前記画像に含まれる認識対象物の輪郭を抽出する動的輪郭抽出方法において、前記画像は認識対象物と背景部とを含む2値画像であり、初期閉曲線を、複数のタイル状に分割してなる各分割領域の周縁に一致するように配置し、前記認識対象物における前記外部エネルギーを前記閉曲線が膨張する方向へ、前記背景部における前記外部エネルギーを前記閉曲線が縮小する方向へ、前記画像の濃度値に応じて動的に変化させるようにしたことから、認識対象物に対して配置する閉曲線の初期設定条件を求めることなく、画像を単純に分割線で小領域に分割し、その領域毎に認識対象物を正確に認識することができ、画像に含まれる認識対象物の認識処理速度を向上させることが可能となる。
【図面の簡単な説明】
【図1】本実施の形態1に係る画像認識方法の動作概要を説明するための図である
【図2】本実施の形態2に係る画像認識方法の動作概要を説明するための図である。
【図3】従来の動的輪郭抽出方法の動作制御を説明するための図である。
【図4】一般的なコンピュータにより画像処理を行う場合のデータフローを示す図である。
【図5】従来の動的輪郭抽出方法において、画像認識処理の処理速度を高速化する際の問題点を説明するための図である。
【符号の説明】
100 2値画像
110,220,310 認識対象物
120,150,320,330 閉曲線
130,131,132,133,134,135,136,137,140,141,330,331,332,333,334,335,336,337,338,339,340,341,342,343 制御点
145 ベクトル
146 外部エネルギー
160 背景部
200,500 画像
210,510 分割線
400 コンピュータシステム
410 キャッシュメモリ
420 CPU
430 メインメモリ
440 ハードディスク[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image recognition method, and more particularly to an energy control method in an active contour extraction method.
[0002]
[Prior art]
As a method for extracting a contour of a specific recognition object included in an image, a dynamic contour extraction method (Snakes: “active contour model”, International Journal of Computer Vision, Vol.1, No.4, pp.321-331). , 1988) is known.
[0003]
In this method, a deformable closed curve is set for a specific recognition object included in an image, and the energy of the closed curve is defined to quantitatively evaluate the state of the closed curve. The energy of the closed curve is defined so as to become the smallest when the closed curve matches the contour of the recognition object. From this, the contour of the recognition object is transformed by deforming the closed curve so that the energy is minimized. Can be extracted. The closed curve is a connection of discretely arranged control points that form the closed curve. Therefore, the process of deforming the above-mentioned closed curve means that each control point on the closed curve is moved so that the energy of the closed curve is minimized. Specifically, the control point v (() on the closed curve is defined so that the sum of the three energies of internal energy (internal spline energy), image energy, and external energy, which is defined corresponding to the state of the closed curve, is minimized. The outline of the recognition object is extracted by moving s). The energy of the closed curve at this time can be expressed by the equation shown in [Formula 1].
[Expression 1]
Figure 0004578638
[0004]
The internal spline energy is an energy that is composed of a first derivative and a second derivative of the control point, and contracts the closed curve and applies a smoothing force, and is expressed by the equation shown in [Equation 2].
[Expression 2]
Figure 0004578638
[0005]
Further, the image energy is given by the differentiation of the image at the control point coordinates, and is an energy that applies a force for sticking the closed curve to the edge of the image, and is expressed by the equation shown in [Equation 3].
[Equation 3]
Figure 0004578638
[0006]
The external energy is energy that gives energy to the closed curve from the outside and applies a force that contracts or expands the closed curve, and is expressed by the equation shown in [Expression 4].
[Expression 4]
Figure 0004578638
[0007]
[Problems to be solved by the invention]
In the conventional active contour extraction method, the closed curve can be deformed only in one direction, either deformation in the reduction direction or deformation in the expansion direction. Therefore, when starting the image recognition process, when performing control for deforming the closed curve in the reduction direction, it is necessary to arrange the closed curve so as to surround the recognition target object. Further, when performing control for deforming the closed curve in the expansion direction, it is necessary to initialize the closed curve so that all line segments are included in the recognition target object. If the initial setting of the closed curve is set so as to intersect with the recognition target area, the closed curve does not stick to the edge of the entire circumference of the recognition target and the recognition fails. That is, as shown in FIG. 3, when the initial closed curve 320 formed by connecting the control points 330 to 343 is set so as to intersect with the recognition object 310, the initial closed curve 320 becomes the closed curve 330 when the reduction control is performed. Due to the deformation, the closed curve 330 does not stick to the edge of the entire circumference of the recognition object 310, and the recognition object 310 cannot be accurately recognized. In other words, the accuracy of recognizing the recognition object in the image in the conventional active contour extraction method largely depends on the initial coordinates of the closed curve, and the recognition object can be recognized depending on the initial state of the closed curve. The possibility of not being possible arises.
[0008]
Further, problems that occur when the processing speed is increased in the image recognition processing by the computer system to which the conventional active contour extraction method is applied will be described below. FIG. 4 shows a data flow when image processing is performed by a general computer system. When performing various types of image processing in the computer system 400, data necessary for image processing is read into the main memory 430 from sensors such as the hard disk 440 and the CCD camera, and the data is transferred to the CPU 420 and then stored in the cache memory 410. Is done. If necessary data has already been transferred to the cache memory 410, the CPU 420 reads the data from the cache memory 410. At this time, the access speed to the cache memory 410 is generally high, while the access speed to the main memory 430 is low. However, since the cache memory 410 is expensive, only a small size is mounted in the computer system. On the other hand, the inexpensive main memory 430 is generally mounted with a large capacity.
[0009]
In order to increase the processing speed of image processing from the above computer structure, it is important to effectively use the cache memory 410 capable of high-speed data access. The ratio indicating that the necessary data of the CPU 420 exists in the cache memory 410 is generically referred to as a hit rate. Improving the hit rate directly leads to a higher speed. As a method for improving the hit rate in a computer system to which the dynamic contour extraction method is applied, the image data is divided into small areas, a closed curve is determined, and the target is recognized for each area, thereby increasing the capacity of the processing target data. It is conceivable to keep the amount small. However, as described above, in the conventional dynamic contour extraction method, when a closed curve is arranged, it is necessary to initially set the closed curve so as not to intersect the recognition target region. However, when the recognition target object 520 is arranged on the image 500 as shown in FIG. 5, the dividing lines 510 that divide the image 500 have a complicated arrangement, which causes a problem that a closed curve cannot be easily determined. Even when the image is divided by the dividing line 510 and the closed curve can be determined, the processing time for solving the dividing line arrangement problem becomes an overhead.
[0010]
Therefore, in the present invention, in the dynamic contour extraction method for extracting the contour of a specific recognition object included in the image, the recognition object is accurately obtained without obtaining the initial setting condition of the closed curve to be arranged for the recognition object. An object of the present invention is to provide an image recognition method that can be recognized easily.
[0011]
[Means for Solving the Problems]
The image recognition method according to claim 1, wherein a closed curve is arranged on the image, and the internal energy indicating the smoothness of the curve, the image energy indicating the density gradient of the image, and external energy from the outside are used. In the dynamic contour extraction method for performing the reduction processing or expansion processing of the closed curve and extracting the contour of the recognition target included in the image , the image is a binary image including the recognition target and a background portion, A closed curve is set so as to intersect with a recognition object, and the external energy in the recognition object is set in a direction in which the closed curve expands, and the external energy in the background portion is set in a direction in which the closed curve reduces. It is characterized by dynamically changing according to the value.
[0012]
The image recognition method according to claim 2, wherein a closed curve is arranged on the image, and internal energy indicating the smoothness of the curve, image energy indicating the density gradient of the image, and external energy from the outside are used. In the dynamic contour extraction method for performing the reduction processing or expansion processing of the closed curve and extracting the contour of the recognition target included in the image, the image is a binary image including the recognition target and a background portion, A closed curve is arranged so as to coincide with the periphery of each divided region obtained by dividing the tile into a plurality of tiles, and the external energy in the recognition target object is expanded in the direction in which the closed curve expands, and the external energy in the background portion is The closed curve is dynamically changed in accordance with the density value of the image in a reducing direction.
DETAILED DESCRIPTION OF THE INVENTION
(Embodiment 1)
Hereinafter, the image recognition method according to the first embodiment will be described with reference to FIG.
In FIG. 1, a binary image 100 includes a recognition target object 110 to be recognized and an outside (background part) 160 of the recognition target object, and the density value (pixel value) of the recognition target object 110 is set to a value of 1, The pixel value of the background portion 160 is set to 0.
[0015]
Hereinafter, a method for recognizing the recognition object 110 in the binary image 100 shown in FIG.
First, as shown in FIG. 1, the initial closed curve 120 is set so as to intersect with the recognition object 110. Here, the closed curve refers to a connection of control points (130 to 137, 140, 141) discretely arranged on the closed curve. In the process of connecting the control points, the direction is assumed to be positive, and the counterclockwise direction is positive. Basic control of the closed curve 120 is performed by moving and controlling each control point so that the energy E SNAKE (v) of the closed curve shown in [Equation 1] is minimized.
[0016]
Next, it is determined whether each control point on the closed curve 120 is located inside or outside the recognition object 110. This determination is performed based on the pixel value of the binary image 100 at each control point coordinate. When the pixel value at the control point coordinate is 1, it is determined that the pixel is located inside the recognition object 110, while the pixel value at the control point coordinate is determined. Is 0, it is determined to be located outside the recognition object 110, that is, in the background portion 160. In FIG. 1, control points outside the recognition target object 110 are indicated by black circles (130 to 137), and control points inside the recognition target object 110 are indicated by white circles (140, 141).
[0017]
Next, the external energy is dynamically changed to determine the moving direction of each control point on the closed curve 120. For control points outside the recognition target object 110, a vector obtained by rotating a vector obtained by connecting control points before and after the target control point by 90 degrees is used as external energy. In addition, a vector obtained by rotating a vector obtained by connecting control points before and after the target control point minus 90 degrees is applied to the control point inside the recognition target object 110 as external energy. Note that a normalized value may be used as the absolute value of the vector.
[0018]
Specifically, for example, when the control point of the movement control target is the control point 131 shown in FIG. 1, the control point 131 exists outside the recognition target object 110, so that the control points are positioned before and after the control point 131. The vector 145 obtained by connecting 130 and the control point 132 is rotated 90 degrees to generate external energy 146. The generated external energy 146 acts on the control point 131 from the outside to the inside of the recognition object 110. Then, the control point 131 moves in accordance with the external energy 146 from the outside to the inside of the recognition object 110, that is, in the direction in which the closed curve 120 is reduced. As described above, external energy is applied to the control points (130 to 137) located outside the recognition target object 110 from the outside to the inside of the recognition target object 110, thereby controlling the control points (130 to 137). Is moved in the direction in which the closed curve 120 is reduced. Further, external energy is applied to the control point (140, 141) located inside the recognition target object 110 from the inside to the outside of the recognition target object 110, whereby the control point (140, 141) is closed on the closed curve 120. Is moved in the direction of expansion. As a result, the initial closed curve 120 is transformed into the closed curve 150 and is attached to the edges of the entire circumference of the recognition object 310, so that the contour of the recognition object 110 is recognized.
[0019]
As described above, according to the image recognition method of the first embodiment, when recognizing the contour of the recognition target object 110 included in the binary image 100, the external energy is moved according to the image value of the binary image 100. Since each control point on the closed curve 120 is moved in a direction in which the closed curve 120 is contracted or expanded according to the external energy, the closed curve 120 is initialized so as to intersect with the recognition object 110. Even in this case, the contour of the recognition object 110 can be recognized accurately.
[0020]
(Embodiment 2)
Hereinafter, an image recognition method according to the second embodiment will be described with reference to FIG.
In the image recognition method according to the second embodiment, when recognizing a specific recognition target included in an image, the image is divided into small regions and recognized for each region division in order to increase the processing speed. It is characterized by recognizing an object.
[0021]
Specifically, first, the image 200 is divided into small regions by the dividing line 210, and the closed curve (231 to 238) is determined. By performing such processing, a plurality of closed curves are arranged in a tile shape with respect to the image 200 as shown in FIG.
[0022]
Next, the recognition target 220 is recognized for each divided region using the image recognition method of the first embodiment. In the second embodiment, since the recognition object 220 is recognized for each divided region using the image recognition method of the first embodiment, a plurality of closed curves are arranged in a tile shape with respect to the image 200. In this case, it is not necessary to perform initial setting so that the recognition target object 220 does not cross the closed curve (231 to 238).
[0023]
As described above, according to the image recognition method of the second embodiment, a plurality of closed curves (231 to 238) are arranged in a tile shape on the image 200, and the recognition target object 220 is imaged in the first embodiment for each region. Since the recognition method is used for recognition, the image 200 is simply divided into tiles and the recognition object 220 is recognized for each area without obtaining the initial setting condition of the closed curve for the recognition object 220. And the recognition processing speed of the recognition object 220 can be improved.
[0024]
【The invention's effect】
As described above, according to the image recognition method of the first aspect, the closed curve is arranged on the image, the internal energy indicating the smoothness of the curve, the image energy indicating the density gradient of the image, and the external external In the dynamic contour extracting method for extracting the contour of the recognition target included in the image by performing reduction processing or expansion processing of the closed curve using energy, the image includes a recognition target and a background portion 2 It is a value image, an initial closed curve is set so as to intersect with a recognition object, the external energy in the recognition object is set in the direction in which the closed curve expands, and the external energy in the background portion is reduced in the direction in which the closed curve reduces. to, since it has to be dynamically changed according to the density value of the image, it is possible to perform the reduction and expansion control of the closed curve corresponding to the pixel values of the image Without obtaining an initial setting condition of closed curves placement against recognition object, it is possible to recognize the contour of the recognition target object included precisely in the image.
[0025]
According to the image recognition method of claim 2, a closed curve is arranged on the image, the internal energy indicating the smoothness of the curve, the image energy indicating the density gradient of the image, and the external energy from the outside In the dynamic contour extraction method for extracting the contour of the recognition object included in the image by performing reduction processing or expansion processing of the closed curve using the image, the image includes a recognition object and a background portion. The initial closed curve is arranged so as to coincide with the peripheral edge of each divided region divided into a plurality of tiles, and the external energy in the recognition object is expanded in the direction of the closed curve in the background portion. the external energy in a direction in which the closed curve is reduced, since it has to be dynamically changed according to the density value of the image, the closed curve placement against recognition object initial Without obtaining a fixed condition, the image can be simply divided into small areas with dividing lines, and the recognition object can be accurately recognized for each area, improving the recognition processing speed of the recognition object included in the image. It becomes possible.
[Brief description of the drawings]
FIG. 1 is a diagram for explaining an operation outline of an image recognition method according to the first embodiment. FIG. 2 is a diagram for explaining an operation outline of an image recognition method according to the second embodiment. .
FIG. 3 is a diagram for explaining operation control of a conventional active contour extraction method.
FIG. 4 is a diagram illustrating a data flow when image processing is performed by a general computer.
FIG. 5 is a diagram for explaining problems in increasing the processing speed of image recognition processing in a conventional active contour extraction method.
[Explanation of symbols]
100 Binary image 110, 220, 310 Recognition object 120, 150, 320, 330 Closed curve 130, 131, 132, 133, 134, 135, 136, 137, 140, 141, 330, 331, 332, 333, 334 335, 336, 337, 338, 339, 340, 341, 342, 343 Control point 145 Vector 146 External energy 160 Background part 200,500 Image 210, 510 Dividing line 400 Computer system 410 Cache memory 420 CPU
430 Main memory 440 Hard disk

Claims (2)

画像上に閉曲線を配置し、曲線の滑らかさを示す内部エネルギーと、前記画像の濃度勾配を示す画像エネルギーと、外部からの外部エネルギーとを用いて、前記閉曲線の縮小処理あるいは膨張処理を行い、前記画像に含まれる認識対象物の輪郭を抽出する動的輪郭抽出方法において、
前記画像は認識対象物と背景部とを含む2値画像であり、
初期閉曲線を認識対象物と交差するように設定し、
前記認識対象物における前記外部エネルギーを前記閉曲線が膨張する方向へ、前記背景部における前記外部エネルギーを前記閉曲線が縮小する方向へ、前記画像の濃度値に応じて動的に変化させることを特徴とする画像認識方法。
A closed curve is arranged on the image, and using the internal energy indicating the smoothness of the curve, the image energy indicating the density gradient of the image, and external energy from the outside, the reduction process or expansion process of the closed curve is performed, In the dynamic contour extraction method for extracting the contour of the recognition object included in the image,
The image is a binary image including a recognition object and a background part,
Set the initial closed curve to intersect the recognition object,
The external energy in the recognition object is dynamically changed in a direction in which the closed curve expands, and the external energy in the background portion is dynamically changed in a direction in which the closed curve is reduced according to the density value of the image. Image recognition method.
画像上に閉曲線を配置し、曲線の滑らかさを示す内部エネルギーと、前記画像の濃度勾配を示す画像エネルギーと、外部からの外部エネルギーとを用いて、前記閉曲線の縮小処理あるいは膨張処理を行い、前記画像に含まれる認識対象物の輪郭を抽出する動的輪郭抽出方法において、
前記画像は認識対象物と背景部とを含む2値画像であり、
初期閉曲線を、複数のタイル状に分割してなる各分割領域の周縁に一致するように配置し、
前記認識対象物における前記外部エネルギーを前記閉曲線が膨張する方向へ、前記背景部における前記外部エネルギーを前記閉曲線が縮小する方向へ、前記画像の濃度値に応じて動的に変化させることを特徴とする画像認識方法。
A closed curve is arranged on the image, and using the internal energy indicating the smoothness of the curve, the image energy indicating the density gradient of the image, and external energy from the outside, the reduction process or expansion process of the closed curve is performed, In the dynamic contour extraction method for extracting the contour of the recognition object included in the image,
The image is a binary image including a recognition object and a background part,
The initial closed curve is arranged so as to coincide with the peripheral edge of each divided area obtained by dividing into a plurality of tiles,
The external energy in the recognition object is dynamically changed in a direction in which the closed curve expands, and the external energy in the background portion is dynamically changed in a direction in which the closed curve is reduced according to the density value of the image. Image recognition method.
JP2000233816A 2000-08-02 2000-08-02 Image recognition method Expired - Fee Related JP4578638B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2000233816A JP4578638B2 (en) 2000-08-02 2000-08-02 Image recognition method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2000233816A JP4578638B2 (en) 2000-08-02 2000-08-02 Image recognition method

Publications (2)

Publication Number Publication Date
JP2002049922A JP2002049922A (en) 2002-02-15
JP4578638B2 true JP4578638B2 (en) 2010-11-10

Family

ID=18726284

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2000233816A Expired - Fee Related JP4578638B2 (en) 2000-08-02 2000-08-02 Image recognition method

Country Status (1)

Country Link
JP (1) JP4578638B2 (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3936683B2 (en) 2003-08-04 2007-06-27 株式会社デンソー Road position detection apparatus and program
JP2006252400A (en) * 2005-03-14 2006-09-21 Keyence Corp Image processor and method of generating registration data in image processing
JP6599697B2 (en) 2015-08-31 2019-10-30 株式会社ミツトヨ Image measuring apparatus and control program therefor

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH10143634A (en) * 1996-11-12 1998-05-29 Fuji Photo Film Co Ltd Method and device for recognizing circular exposure field
JP2000048212A (en) * 1998-07-31 2000-02-18 Canon Inc Device and method for picture processing and recording medium

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH10143634A (en) * 1996-11-12 1998-05-29 Fuji Photo Film Co Ltd Method and device for recognizing circular exposure field
JP2000048212A (en) * 1998-07-31 2000-02-18 Canon Inc Device and method for picture processing and recording medium

Also Published As

Publication number Publication date
JP2002049922A (en) 2002-02-15

Similar Documents

Publication Publication Date Title
CN110175558B (en) Face key point detection method and device, computing equipment and storage medium
WO2021047684A1 (en) Active contour- and deep learning-based automatic segmentation method for fuzzy boundary image
JP4340968B2 (en) Image processing apparatus and method, recording medium, and program
JP5336995B2 (en) Feature point positioning device, image recognition device, processing method thereof, and program
WO2009113231A1 (en) Image processing device and image processing method
JP2008257713A (en) Correcting device and method for perspective transformed document image
JP2008299627A (en) Information processing method and apparatus, program, storage medium
JP2001357399A (en) Device and method for processing picture
JP6364837B2 (en) Image processing apparatus and region dividing method
WO2019222889A1 (en) Image feature extraction method and device
WO2019041967A1 (en) Hand detection method and system, image detection method and system, hand segmentation method, storage medium, and device
JP4409072B2 (en) Outline extraction device, outline extraction method, and recording medium recording outline extraction program
WO2022143284A1 (en) Method and apparatus for determining movement direction, and sweeping robot and storage medium
JP4578638B2 (en) Image recognition method
CN115761826A (en) Palm vein effective area extraction method, system, medium and electronic device
JP3054682B2 (en) Image processing method
CN113379806A (en) Target tracking method and system based on learnable sparse conversion attention mechanism
JP2015026117A (en) Image processing method, image processing apparatus, program, and recording medium
WO2018053710A1 (en) Morphological processing method of digital image and digital image processing device
JP3638845B2 (en) Image processing apparatus and method
CN107798684A (en) A kind of active contour image partition method and device based on symbol pressure function
JP2001082927A (en) Method and apparatus for processing three-dimensional image and recording medium recording three-dimensional image processing program
WO2020255645A1 (en) Three-dimensional data updating device, face orientation estimation device, three-dimensional data updating method, and computer-readable recording medium
JP2010009437A (en) Matching system corresponding to disappearance of pattern and inspection device using the same
TWI815351B (en) Target tracking method and target tracking device

Legal Events

Date Code Title Description
A621 Written request for application examination

Free format text: JAPANESE INTERMEDIATE CODE: A621

Effective date: 20070621

A977 Report on retrieval

Free format text: JAPANESE INTERMEDIATE CODE: A971007

Effective date: 20100416

A131 Notification of reasons for refusal

Free format text: JAPANESE INTERMEDIATE CODE: A131

Effective date: 20100511

A521 Written amendment

Free format text: JAPANESE INTERMEDIATE CODE: A523

Effective date: 20100709

TRDD Decision of grant or rejection written
A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

Effective date: 20100727

A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

A61 First payment of annual fees (during grant procedure)

Free format text: JAPANESE INTERMEDIATE CODE: A61

Effective date: 20100825

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20130903

Year of fee payment: 3

R150 Certificate of patent or registration of utility model

Free format text: JAPANESE INTERMEDIATE CODE: R150

S111 Request for change of ownership or part of ownership

Free format text: JAPANESE INTERMEDIATE CODE: R313113

R350 Written notification of registration of transfer

Free format text: JAPANESE INTERMEDIATE CODE: R350

LAPS Cancellation because of no payment of annual fees