この発明に係る画像処理装置は、入力された画像から複数のズレ量検出対象範囲を抽出するズレ量検出対象範囲抽出手段と、複数のズレ量検出対象範囲内の一方向毎に画素値を積算して、積算値を算出する画素値積算手段と、複数のズレ量検出対象範囲毎に、積算値を用いて、当該ズレ量検出対象範囲における一方向と異なる方向である別方向の変位特徴量を算出する変位特徴量算出手段と、画像より先に入力された画像内の所定のズレ量検出対象範囲であるマッチングブロックと複数のズレ量検出対象範囲との間の変位特徴量の相関を演算し相関指標値を算出する相関指標値算出手段とを設け、ズレ量検出手段が、複数の相関指標値の中で最も相関が高い旨を示している相関指標値に対応するズレ量検出対象範囲を特定し、そのズレ量検出対象範囲とマッチングブロックとの別方向のズレ量を検出するようにしたものである。
The image processing apparatus according to the present invention, a shift amount detection target range extracting means for extracting a plurality of shift amount detection target range from the input image, the pixel values for each direction in the shift amount detection target range of several integrating to another and image pixel value integrating means you calculate a totalized value, for each of a plurality of shift amount detection target range, using a totalized value, a one-way direction different in the shift amount detection target range between the displacement of the feature amount calculating means that to calculate the displacement characteristic of a direction, a matching block and a plurality of shift amount detection target range is a predetermined shift amount detection target range in an image that is input earlier than images the correlation of the displacement characteristic quantity calculated provided the correlation index value calculating means that to calculate the correlation index value,'s record amount detecting means, most indicates that high correlation among the correlation index value of multiple Detect the deviation detection target range corresponding to the correlation index value and detect the deviation It is obtained to detect the different directions of the shift amount between elephant range and the matching block.
この発明によれば、入力された画像から複数のズレ量検出対象範囲を抽出するズレ量検出対象範囲抽出手段と、複数のズレ量検出対象範囲内の一方向毎に画素値を積算して、積算値を算出する画素値積算手段と、複数のズレ量検出対象範囲毎に、積算値を用いて、当該ズレ量検出対象範囲における一方向と異なる方向である別方向の変位特徴量を算出する変位特徴量算出手段と、画像より先に入力された画像内の所定のズレ量検出対象範囲であるマッチングブロックと複数のズレ量検出対象範囲との間の変位特徴量の相関を演算し相関指標値を算出する相関指標値算出手段とを設け、ズレ量検出手段が、複数の相関指標値の中で最も相関が高い旨を示している相関指標値に対応するズレ量検出対象範囲を特定し、そのズレ量検出対象範囲とマッチングブロックとの別方向のズレ量を検出するように構成したので、別方向に大きな画像ズレが発生しても、２枚の画像間のズレ量を高精度に検出することができる効果がある。
According to the present invention, a shift amount detection target range extracting means for extracting a plurality of shift amount detection target range from the input image, by integrating the pixel values in one direction each in the shift amount detection target range of several the image pixel value integrating means you calculate a totalized value, for each of a plurality of shift amount detection target range, using a totalized value, different directions of displacement in one direction different from the direction in the shift amount detection target range displacement characteristics between the displacement of the feature amount calculating means that to calculate the feature amount, and the matching block and a plurality of shift amount detection target range is a predetermined shift amount detection target range in an image that is input earlier than images a correlation index value calculating means that to calculate the correlation index value to calculate the correlation amount provided, correlation's record amount detecting means indicates a fact most highly correlated in the correlation index value of multiple The deviation amount detection target range corresponding to the index value is specified, and the deviation amount detection target range and the map are detected. Since it is configured to detect different directions of displacement of the quenching blocks, even if a large image shift to a different direction is generated, there is an effect that it is possible to detect the shift amount between the two images with high precision .