JPH0410874A - Movement vector detecting device - Google Patents
Movement vector detecting deviceInfo
- Publication number
- JPH0410874A JPH0410874A JP2113955A JP11395590A JPH0410874A JP H0410874 A JPH0410874 A JP H0410874A JP 2113955 A JP2113955 A JP 2113955A JP 11395590 A JP11395590 A JP 11395590A JP H0410874 A JPH0410874 A JP H0410874A
- Authority
- JP
- Japan
- Prior art keywords
- output
- subtracter
- classification
- averaging
- circuit
- 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.)
- Pending
Links
- 239000013598 vector Substances 0.000 title claims abstract description 21
- 238000012935 Averaging Methods 0.000 claims abstract description 12
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000001514 detection method Methods 0.000 claims description 13
- 238000011156 evaluation Methods 0.000 abstract description 2
- 238000000034 method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 2
- 238000006073 displacement reaction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
Landscapes
- Studio Circuits (AREA)
- Closed-Circuit Television Systems (AREA)
- Color Television Systems (AREA)
- Processing Of Color Television Signals (AREA)
Abstract
Description
【発明の詳細な説明】
［産業上の利用分野コ
本発明は動きベクトル検出装置に関し、より具体的には
、ＴＶカメラ、電子カメラ、ビデオ・カメラ及び工業用
画像計測機器などの撮像光学装置、特に防振や追尾など
の機能を有する撮像光学装置における動きベクトル検出
装置に関する。DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a motion vector detection device, and more specifically, to an imaging optical device such as a TV camera, an electronic camera, a video camera, and an industrial image measuring device; In particular, the present invention relates to a motion vector detection device in an imaging optical device having functions such as image stabilization and tracking.
［従来の技術］
画像信号処理による動きベクトル検出法としては、米国
特許第３８９０４６２号明細書、昭和６０年特許出願公
告第４６８７８号やＪ、Ｏ，ＬＬｍｂ ａｎｄ Ｊ、Ａ
、ＭｕｒｐｈｙＭｅａｓｕｒｉｎｇ ｔｈｅ ５ｐｅｅ
ｄ ｏｆ Ｍｏｖｉｎｇ ０ｂｊｅｃｔｓ ｆｒｏｍ
ＴｅＬｅｖｉｓｉｏｎ Ｓｉｇｎａｌｓ”、ＩＥＥＥ
Ｔｒａｎｓ、 Ｃｏｍ、、Ｃｏｍ−２３，４ｐｐ４７４
−４７８（Ａｐｒｉｌ １９７５）等に記載される時空
間勾配法がある。この時空間勾配法では、Ｘ方向の画像
変位量をα、Ｘ方向の画像変位量をβとすると、α＝Σ
＠ ｄ−ｓｉｇｎ＜ｇｙ’）／Σｗｉｇｘ’β＝ΣＢｄ
−ｓｉｇｎ（ｇｙ’）／Σｓ ｌ ｇｙ’ １で与えら
れる。但し、ｄは時間的に連続する画像間の同じ位置に
おける濃度（レベル）差、即ち時間勾配を示し、ｇ−’
９ｇｙ°はそれぞれ画像をｇで表わしたときのＸ方向、
Ｘ方向の空間勾配を示す。[Prior Art] As a motion vector detection method using image signal processing, there are disclosed US Pat.
,MurphyMeasuring the 5pee
d of Moving 0 objects from
TeLevision Signals”, IEEE
Trans, Com,, Com-23,4pp474
-478 (April 1975), etc., is a spatiotemporal gradient method. In this spatio-temporal gradient method, if the amount of image displacement in the X direction is α and the amount of image displacement in the X direction is β, then α=Σ
@d-sign<gy')/Σwigx'β=ΣBd
−sign(gy')/Σs l gy'1. However, d indicates the density (level) difference at the same position between temporally consecutive images, that is, the temporal gradient, and g-'
9gy° is the X direction when the image is expressed in g,
The spatial gradient in the X direction is shown.
また、Σ、は、複数個の画素からなる単位演算領域であ
るブロック毎の総和演算を意味し、ｓｉｇｎ０はｇ８°
＊ ｇｙ’の符号を出力する関数である。In addition, Σ means a summation calculation for each block, which is a unit calculation area consisting of a plurality of pixels, and sign0 is g8°
* This is a function that outputs the sign of gy'.
［発明が解決しようとする課題］
しかし、上記従来例では、Ｘ方向の空間勾配と濃度差か
らＸ方向の動き量を求め、Ｘ方向の空間勾配と濃度差か
らＸ方向の動き量を求めており、ｘ、 ｙ相互の方向
の空間勾配の影響を考慮していないので、画像の二次元
的な動きに対応できないという欠点がある。[Problems to be Solved by the Invention] However, in the above conventional example, the amount of movement in the X direction is determined from the spatial gradient in the X direction and the density difference, and the amount of movement in the X direction is determined from the spatial gradient in the X direction and the density difference. However, since it does not take into account the influence of spatial gradients in the x and y directions, it has the disadvantage that it cannot cope with two-dimensional movement of the image.
第２図を参照して、より具体的に説明する。第２図は、
斜め（右下）方向の勾配を持つ画像を１６個のブロック
に分割し、第２図上部に示した右上方向の動きを与えた
場合の、従来例による検出結果をブロック毎の動きベク
トルで表示したものである。第２図の円弧は等濃度線で
ある。This will be explained in more detail with reference to FIG. Figure 2 shows
When an image with a gradient in the diagonal (lower right) direction is divided into 16 blocks and motion is applied in the upper right direction as shown in the upper part of Figure 2, the detection results using the conventional example are displayed as motion vectors for each block. This is what I did. The arcs in FIG. 2 are isodensity lines.
ブロック＃６．＃１１．＃１６では、ブロック内に斜め
方向の空間勾配成分が強い。ところが、時空間勾配法で
は、「ブロック内には斜め方向の大きな空間勾配が無い
」ことを大前提としており、これに反するので、ｘ、７
両方向ともに正しい推定が得られていない。Block #6. #11. In #16, there is a strong spatial gradient component in the diagonal direction within the block. However, the spatio-temporal gradient method assumes that there is no large spatial gradient in the diagonal direction within the block, which is contrary to the assumption that x, 7
Correct estimates are not obtained in both directions.
また、ブロック＃２．＃３．＃４では、Ｘ方向の動きは
正しく推定しているものの、Ｘ方向の動きは正しく推定
していない。ブロック＃５．＃９゜＃１３では、ブロッ
ク＃２．＃３．＃４と逆に、Ｘ方向の動きは正しく推定
しているものの、Ｘ方向の動きは正しく推定していない
。これは、時空間勾配法では、「空間勾配が無い方向の
動きは推定できない」からである。また、ブロック＃７
゜＃８．＃１０．＃１４では、時空間勾配法の同様の原
理により、空間勾配の小さい方向で大きなエラーが発生
している。Also, block #2. #3. In #4, although the motion in the X direction is estimated correctly, the motion in the X direction is not estimated correctly. Block #5. In #9° #13, block #2. #3. Contrary to #4, although the motion in the X direction is estimated correctly, the motion in the X direction is not. This is because the spatio-temporal gradient method "cannot estimate movement in a direction where there is no spatial gradient." Also, block #7
゜#8. #10. In #14, a large error occurs in a direction with a small spatial gradient due to the same principle of the spatiotemporal gradient method.
そこで本発明は、これらの問題点を解決し、実時間処理
に対応できる高速で高精度な動きベクトル検出装置を提
示することを目的としている。SUMMARY OF THE INVENTION An object of the present invention is to solve these problems and provide a high-speed, high-precision motion vector detection device that can handle real-time processing.
［課題を解決するための手段］
本発明に係る動きベクトル検出装置は、画面間の濃度差
及び画面内の空間勾配から画像の動き量を求める動きベ
クトル検出装置であって、斜め方向空間勾配成分の影響
を近似した推定式に従いブロック毎の動きベクトルを演
算する第１の演算手段と、入力画像信号の斜め方向空間
勾配成分量を表わす信号を求める第２の演算手段と、第
２の演算手段の演算結果に応じたパターンに各ブロック
を分類する分類手段と、当該分類手段の分類に応じた重
みもとで当該第１の演算手段による各ブロックの動きベ
クトルを平均化する平均化手段とからなることを特徴と
する。[Means for Solving the Problems] A motion vector detection device according to the present invention is a motion vector detection device that calculates the amount of movement of an image from the density difference between screens and the spatial gradient within the screen, and which calculates the amount of movement of an image from the density difference between screens and the spatial gradient within the screen. a first calculation means for calculating a motion vector for each block according to an estimation formula that approximates the influence of the input image signal; a second calculation means for calculating a signal representing the amount of the diagonal spatial gradient component of the input image signal; a classification means for classifying each block into a pattern according to the calculation result of the first calculation means; and an averaging means for averaging the motion vectors of each block by the first calculation means under a weight according to the classification of the classification means. It is characterized by becoming.
［作用コ
上記第１の演算手段により、基本的には近似推定式によ
り動きベクトルを求めているので、高速である。また、
上記第２の演算手段及び分類手段により斜め方向空間勾
配成分量を考慮して各ブロックを分類し、平均化手段か
その分類に応じた重みで平均化を行なうので、近似推定
式に起因する誤差を有する推定結果の評価を低くでき、
従って、全体として、高い信頼性で動きベクトルを求め
ることができる。[Operations] The first calculation means basically calculates the motion vector using an approximate estimation formula, so it is fast. Also,
Since each block is classified by the above-mentioned second calculation means and classification means in consideration of the amount of diagonal spatial gradient components, and the averaging means performs averaging using weights according to the classification, errors caused by the approximate estimation formula It is possible to lower the evaluation of estimation results with
Therefore, overall, motion vectors can be determined with high reliability.
［実施例コ 以下、図面を参照して本発明の詳細な説明する。[Example code] Hereinafter, the present invention will be described in detail with reference to the drawings.
第１図は、テレビジョン・カメラやビデオ・カメラなど
の画面揺れ防止装置に適用した本発明の一実施例の構成
ブロック図を示す。１０は画像信号の入力端子、１２は
１フイールド（又はフレーム）期間、入力信号を記憶（
即ち、時間遅延）するレジスタ、１４は入力端子１０の
画像信号からレジスタ１２の出力を減算する減算器、１
６は画像濃度分布のＸ方向空間勾配を演算するのに必要
な数画素分の走査時間だけ、入力信号を記憶（即ち、時
間遅延）するレジスタ、１８は入力端子１０の画像信号
からレジスタ１６の出力を減算する減算器、２０はＸ方
向の空間勾配を求めるために、所定走査期間だけ入力端
子１０の画像信号を記憶（即ち、時間遅延）するレジス
タ、２２は入力端子１０の画像信号からレジスタ２０の
出力を減算する減算器である。減算器１４の出力は濃度
勾配ｄであり、減算器１８の出力はＸ方向の空間勾配ｇ
、°、減算器２２の出力はＸ方向の空間勾配ｇｙ’であ
る。FIG. 1 shows a block diagram of an embodiment of the present invention applied to a screen shake prevention device for a television camera, a video camera, or the like. 10 is an input terminal for image signals; 12 is a terminal for storing input signals for one field (or frame) period;
14 is a subtracter for subtracting the output of the register 12 from the image signal at the input terminal 10;
Reference numeral 6 indicates a register for storing (i.e., time delay) the input signal for the scanning time of several pixels necessary to calculate the X-direction spatial gradient of the image density distribution. A subtractor 20 subtracts the output, a register 20 stores the image signal of the input terminal 10 for a predetermined scanning period (i.e., time delay) in order to obtain the spatial gradient in the X direction, and a register 22 stores the image signal of the input terminal 10. This is a subtracter that subtracts the output of 20. The output of the subtractor 14 is the concentration gradient d, and the output of the subtractor 18 is the spatial gradient g in the X direction.
,°, the output of the subtractor 22 is the spatial gradient gy' in the X direction.
２４．２６はそれぞれ減算器１ｓ； ２２の出力の符
号（正、負又はゼロ）を示す信号を出力する符号出力回
路、２８．３０はそれぞれ、符号出力回路２４．２６の
出力に減算器１４の出力（濃度勾配ｄ）を乗算する乗算
器、３２．３４はそれぞれ、減算器１８．２２の出力の
絶対値を出力する絶対値回路、３６は減算器１８の出力
ｇｘ′に減算器２２の出力ｇｙ°を乗算する乗算器、３
８，４０４２．４４．４６，４８．５０は指定ブロック
内で、それぞれ乗算器２８、絶対値回路３２、乗算器３
０、絶対値回路３４、乗算器３６、減算器１８及び減算
器２２の出力データを累積加算する総和回路、５２は総
和回路３８の出力を総和回路４０の出力で除算する除算
器、５４は総和回路４２の出力を総和回路４４の出力で
除算する除算器、５６は総和回路４８の出力を総和回路
５０の出力で除算する除算器である。除算器５２の出力
はＸ方向の動き量を示し、除算器５４の出力はＸ方向の
動き量を示す。５８は除算器５６の出力の絶対値を出力
する絶対値回路である。24 and 26 are each a subtracter 1s; a sign output circuit that outputs a signal indicating the sign (positive, negative, or zero) of the output of 22; 28 and 30 are each a subtracter 1s that outputs a signal indicating the sign (positive, negative, or zero) of the output of the sign output circuit 24 and 26; A multiplier that multiplies the output (concentration gradient d), 32 and 34 are absolute value circuits that output the absolute values of the outputs of the subtracters 18 and 22, respectively, and 36 is the output gx' of the subtracter 18 and the output of the subtracter 22. Multiplier that multiplies gy°, 3
8,4042.44.46 and 48.50 are the multiplier 28, absolute value circuit 32, and multiplier 3 within the designated block, respectively.
0, a summation circuit that cumulatively adds the output data of the absolute value circuit 34, the multiplier 36, the subtracter 18, and the subtracter 22; 52, a divider that divides the output of the summation circuit 38 by the output of the summation circuit 40; 54, a summation circuit; A divider 56 divides the output of the circuit 42 by the output of the summation circuit 44, and a divider 56 divides the output of the summation circuit 48 by the output of the summation circuit 50. The output of the divider 52 indicates the amount of motion in the X direction, and the output of the divider 54 indicates the amount of motion in the X direction. 58 is an absolute value circuit that outputs the absolute value of the output of the divider 56.
６０は除算器５２．５４から出力される動き量、並びに
、総和回路４６及び絶対値回路５８から出力される空間
勾配情報を記憶するメモリ、６２は、詳細は後述するが
、メモリ６０に記憶される当該空間勾配情報に従い、所
定の複数の空間勾配パターンに各ブロックを分類するパ
ターン分類回路、６４は分類回路６２により分類された
ブロックに最適な重みを付けて平均化する平均化回路、
６６は検出結果の動きベクトルを出力する出力端子であ
る。A memory 60 stores the amount of motion output from the dividers 52 and 54 and spatial gradient information output from the summation circuit 46 and the absolute value circuit 58; a pattern classification circuit that classifies each block into a plurality of predetermined spatial gradient patterns according to the spatial gradient information; 64 is an averaging circuit that applies optimal weights to the blocks classified by the classification circuit 62 and averages them;
66 is an output terminal that outputs the motion vector of the detection result.
次に、第１図の動作を説明する。レジスタ１２及び減算
器１４は、入力端子１０に入力した画像信号ｇから、時
間的に連続する２つのフィールド（又はフレーム）画面
間での濃度差、即ち時間勾配ｄを算出し、レジスタ１６
及び減算器１８は、前フィールド（又はフレーム）画面
内のＸ方向の空間勾配ｇ８°を算出し、レジスタ２０及
び減算器２２は、前フィールド（又はフレーム）画面内
のＸ方向の空間勾配ｇｙ’を算出する。Next, the operation shown in FIG. 1 will be explained. The register 12 and the subtracter 14 calculate the density difference between two temporally consecutive field (or frame) screens, that is, the temporal gradient d, from the image signal g input to the input terminal 10.
and the subtractor 18 calculates the spatial gradient g8° in the X direction in the previous field (or frame) screen, and the register 20 and the subtractor 22 calculate the spatial gradient gy' in the X direction in the previous field (or frame) screen. Calculate.
符号出力回路２４は減算器１８の出力ｇ、′が正のとき
には、＋１、セロのときには０１負のときには−１を出
力し、絶対値回路３０は空間勾配ｇ０゛の絶対値を出力
し、乗算器２８は減算器１４の出力ｄに符号出力回路２
４の出力を乗算する。同様に、符号出力回路２６は減算
器２２の出力ｇ。The sign output circuit 24 outputs +1 when the output g,' of the subtracter 18 is positive, 0 when it is zero, and -1 when it is negative, and the absolute value circuit 30 outputs the absolute value of the spatial gradient g0'', and multiplies it. The subtracter 28 outputs the sign output circuit 2 to the output d of the subtracter 14.
Multiply the output of 4. Similarly, the sign output circuit 26 receives the output g of the subtracter 22.
の符号を示す信号を出力し、絶対値回路３４は空間勾配
ｇｙ’の絶対値を出力し、乗算器３０は減算器１４の出
力ｄに符号出力回路２６の出力を乗算する。そして、総
和回路３８，４０，４２．４４はそれぞれ、所定数の画
素からなるブロック内で、乗算器２８の出力、絶対値回
路３２の出力、乗算器３０の出力、及び絶対値回路３４
の出力の総和を求める。即ち、総和回路３８はΣ、 ｄ
−ｓｉｇｎ（ｇｏ’）を、総和回路４０はΣ、１訃°１
を、総和回路４２はΣ、 ｄ−ｓｉｇｎ（ｇｙ’）を、
総和回路４４はΣａ ｌ ｇｙを計算する。除算器５２
は総和回路３８の出力を総和回Ｉｌ！ｆ！Ｉ４０の出力
で除算し、除算器５４は総和回路４２の出力を総和回路
４４の出力で除算する。除算器５２．５４の出力はそれ
ぞれ、上式のα、βに相当する。除算器５２．５４の出
力はメモリ６０に一時格納される。The absolute value circuit 34 outputs the absolute value of the spatial gradient gy', and the multiplier 30 multiplies the output d of the subtracter 14 by the output of the sign output circuit 26. The summation circuits 38, 40, 42, and 44 each output the output of the multiplier 28, the output of the absolute value circuit 32, the output of the multiplier 30, and the absolute value circuit 34 within a block consisting of a predetermined number of pixels.
Find the sum of the outputs. That is, the summation circuit 38 has Σ, d
-sign(go'), the summation circuit 40 is Σ, 1 °1
, the summation circuit 42 calculates Σ, d-sign(gy'),
A summation circuit 44 calculates Σa l gy. Divider 52
is the output of the summation circuit 38 as the summation circuit Il! f! The divider 54 divides the output of the summation circuit 42 by the output of the summation circuit 44. The outputs of the dividers 52 and 54 correspond to α and β in the above equation, respectively. The outputs of dividers 52 and 54 are temporarily stored in memory 60.
また、乗算器３６は減算器１８の出力ｇ８°に減算器２
２の出力ｇｙ’を乗算し、総和回路４６は所定ブロック
内で、乗算器３６の出力の総和を計算する。総和回路４
８．５０はそれぞれ、所定ブロック内で減算器１８．２
２の出力ｇ、’、 ｇ、’の総和を計算し、除算器５６
は総和回路４８の出力を総和回路５０の出力で除算し、
絶対値回路５８は除算器５６の出力の絶対値を出力する
。即ち、絶対値回路５８は、１ΣＢｇ−’／Σｉｇｙ’
ｌを出力する。Furthermore, the multiplier 36 outputs the subtracter 2 to the output g8° of the subtracter 18.
The summation circuit 46 calculates the sum of the outputs of the multipliers 36 within a predetermined block. Summation circuit 4
8.50 each subtractor 18.2 within a given block.
The sum of the outputs g,', g,' of 2 is calculated, and the divider 56
divides the output of the summation circuit 48 by the output of the summation circuit 50,
Absolute value circuit 58 outputs the absolute value of the output of divider 56. That is, the absolute value circuit 58 calculates 1ΣBg-'/Σigy'
Output l.
総和回路４６の出力Σｌ１ｇ。°・ｇｙ’と、絶対値回
路５８の出力はメモリ６０に供給される。Output Σl1g of summation circuit 46. °·gy' and the output of the absolute value circuit 58 are supplied to a memory 60.
パターン分類回路６２はメモリ６０に記憶される１Σａ
ｇ、’／Σａｇｙ’ｌとΣ９ｇ８゛・ｇｙ′の情報から
、各ブロックを複数の単純なパターンに分類する。The pattern classification circuit 62 stores 1Σa in the memory 60.
From the information of g,'/Σagy'l and Σ9g8'·gy', each block is classified into a plurality of simple patterns.
その分類動作のフローチャートを第３図に図示した。即
ち、ΣＢｇｘ°・ｇ、“か所定の閾値Ｔｈより大きい場
合には（Ｓｌ）、それは斜め方向の空間勾配成分が大き
いと判断して、このブロックの出力結果を除外する（Ｓ
２）。単純に除外せずに、Σ６ｇ、°・ｇｙ′に応じた
重ろを付けるようにしてもよい。次に、１Σｓｇ−’／
Σｍｇｙ’ｌから、ＩΣおｇ８°１がＩΣａｇｙ’ｌよ
り充分に大きい場合（Ｓ３）、Ｘ方向の推定結果αをよ
り重く、Ｘ方向の推定結果βをより軽く評価しくＳ４）
、１Σｍｇｙ°１が１Σｍｇｙ’ｌと同程度の場合には
ｘ、ｙ両方の推定結果α、βを共に信頼性か高いとして
重く評価しくＳ５）、ｌΣＢｇ−’ ｌが１ΣＢｇｙよ
り充分に小さい場合には、Ｘ方向の推定結果βをより重
く、Ｘ方向の推定結果αをより軽く評価する（Ｓ６）。A flowchart of the classification operation is shown in FIG. That is, if ΣBgx°·g is larger than a predetermined threshold Th (Sl), it is determined that the spatial gradient component in the diagonal direction is large, and the output result of this block is excluded (S
2). Instead of simply excluding them, weights may be given according to Σ6g and °·gy'. Next, 1Σsg−'/
From Σmgy'l, if IΣog8°1 is sufficiently larger than IΣagy'l (S3), evaluate the estimation result α in the X direction more heavily and the estimation result β in the X direction lighter S4)
, 1Σmgy°1 is comparable to 1Σmgy'l, the estimation results α and β of both x and y are highly evaluated as reliable or highS5), and when lΣBg-'l is sufficiently smaller than 1ΣBgy' evaluates the estimation result β in the X direction more heavily and the estimation result α in the X direction less heavily (S6).
このように各プロ・レフを分類し、平均化回路６４はそ
の一分類に従った重みを付けて各ブロックの推定結果を
平均化し、出力端子６６から出力する。In this way, each pro-ref is classified, and the averaging circuit 64 averages the estimation results of each block by weighting them according to the classification, and outputs the result from the output terminal 66.
［発明の効果］
以上の説明から容易に理解できるように、本発明によれ
ば、空間勾配の無い方向の動きに対する検出エラーを抑
制でき、また、ブロック内に大きな斜め方向の空間勾配
かあることによる検出誤差も抑制できる。従って、動き
ベクトルを高速且つ高精度に検出できる。[Effects of the Invention] As can be easily understood from the above explanation, according to the present invention, detection errors for movements in directions with no spatial gradient can be suppressed, and detection errors can be suppressed even when there is a large spatial gradient in a diagonal direction within a block. Detection errors caused by this can also be suppressed. Therefore, motion vectors can be detected at high speed and with high precision.
第１図は本発明の一実施例の構成ブロック図、第２図は
従来例ではシミュレーション結果、第３図は本実施例に
よるパターン分類のフローチャートである。
１０：入力端子 １２，１６．２０・レジスタ１４．１
８，２２：減算器 ２４，２６：符号出力回路 ２８，
３０，３６：乗算器 ３２，３４５８：絶対値回路 ３
８，４０，４２，４４，４６．４８，５０：総和回路
５２，５４，５６：除算器 ６０：メモリ ６２：パタ
ーン分類回路６４：平均化回路 ６６：出力端子
第２図
第１図
第３図FIG. 1 is a configuration block diagram of an embodiment of the present invention, FIG. 2 is a simulation result of a conventional example, and FIG. 3 is a flowchart of pattern classification according to this embodiment. 10: Input terminal 12, 16.20・Register 14.1
8, 22: Subtractor 24, 26: Sign output circuit 28,
30, 36: Multiplier 32, 3458: Absolute value circuit 3
8, 40, 42, 44, 46. 48, 50: Summation circuit
52, 54, 56: Divider 60: Memory 62: Pattern classification circuit 64: Averaging circuit 66: Output terminal Figure 2 Figure 1 Figure 3
Claims (1)
を求める動きベクトル検出装置であって、斜め方向空間
勾配成分の影響を近似した推定式に従いブロック毎の動
きベクトルを演算する第１の演算手段と、入力画像信号
の斜め方向空間勾配成分量を表わす信号を求める第２の
演算手段と、第２の演算手段の演算結果に応じたパター
ンに各ブロックを分類する分類手段と、当該分類手段の
分類に応じた重みもとで当該第１の演算手段による各ブ
ロックの動きベクトルを平均化する平均化手段とからな
ることを特徴とする動きベクトル検出装置。A motion vector detection device that calculates the amount of image movement from the density difference between screens and the spatial gradient within the screen, the first device calculating a motion vector for each block according to an estimation formula that approximates the influence of the diagonal spatial gradient component. a calculation means, a second calculation means for obtaining a signal representing the amount of the diagonal spatial gradient component of the input image signal, a classification means for classifying each block into a pattern according to the calculation result of the second calculation means, and said classification. 1. A motion vector detection device comprising: averaging means for averaging the motion vectors of each block produced by the first calculating means under a weight according to the classification of the means.
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2113955A JPH0410874A (en) | 1990-04-27 | 1990-04-27 | Movement vector detecting device |
| EP91303792A EP0454481B1 (en) | 1990-04-27 | 1991-04-26 | Movement vector detection device |
| DE69121627T DE69121627T2 (en) | 1990-04-27 | 1991-04-26 | Device for the detection of motion vector |
| US07/967,569 US5296925A (en) | 1990-04-27 | 1992-10-27 | Movement vector detection device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2113955A JPH0410874A (en) | 1990-04-27 | 1990-04-27 | Movement vector detecting device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| JPH0410874A true JPH0410874A (en) | 1992-01-16 |
Family
ID=14625397
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP2113955A Pending JPH0410874A (en) | 1990-04-27 | 1990-04-27 | Movement vector detecting device |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JPH0410874A (en) |
-
1990
- 1990-04-27 JP JP2113955A patent/JPH0410874A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Odobez et al. | Robust multiresolution estimation of parametric motion models | |
| KR930002613B1 (en) | Image motion vector detector | |
| US8416993B2 (en) | Object boundary accurate motion detection using hierarchical block splitting and motion segmentation | |
| US7949205B2 (en) | Image processing unit with fall-back | |
| US20180144485A1 (en) | Image processing apparatus, image processing method, and non-transitory computer-readable storage medium | |
| US7499494B2 (en) | Vector selection decision for pixel interpolation | |
| CN106791279B (en) | Motion compensation method and system based on occlusion detection | |
| JP2000306108A (en) | Optical flow estimation method | |
| US5453800A (en) | Apparatus for judging a hand movement of an image | |
| US8817869B2 (en) | Image processing device and method, and image display device and method | |
| EP0586708B1 (en) | Image processor and method therefor | |
| JPH0787481A (en) | Fault detector | |
| EP0383245B1 (en) | Apparatus for detecting moving and unmoving regions | |
| JPH0220988A (en) | Moving vector detection system for animation encoding device | |
| JPH0410874A (en) | Movement vector detecting device | |
| Ben-Ezra et al. | Motion segmentation using convergence properties | |
| JP4289170B2 (en) | Noise amount measuring apparatus and video receiver | |
| JPH0410885A (en) | Movement vector detecting device | |
| JPH03140070A (en) | Moving vector detector | |
| JPH0562876B2 (en) | ||
| JPH0410872A (en) | Movement vector detecting device | |
| Giaccone et al. | Motion-compensated multichannel noise reduction of color film sequences | |
| JP2975146B2 (en) | Motion vector detection device | |
| JP3003044B2 (en) | Apparatus and method for detecting motion vector | |
| JPH08223540A (en) | Motion interpolation method using motion vector, motion interpolation circuit, motion vector detection method and motion vector detection circuit |