使用OpenCV 实现matlab的padarray（A, padsize, ‘symmetric’）函数简单实现

最新推荐文章于 2021-04-15 19:43:38 发布

原创最新推荐文章于 2021-04-15 19:43:38 发布 · 2.8k 阅读

3 ·

CC 4.0 BY-SA版权

文章标签：

#opencv #matlab #padarray

opencv 专栏收录该内容

3 篇文章

订阅专栏

本文介绍Matlab中padarray方法用于数组边界的镜像填充，并提供了使用OpenCV C++实现的详细代码示例。通过具体示例展示如何在向量和矩阵周围添加镜像填充。

Matlab文档padarray方法说明：
B = padarray(A,padsize,padval) pads array A where padval specifies the value to use as the pad value. padarray uses the value 0 (zero) as the default. padval can be a scalar that specifies the pad value directly or one of the following text strings that specifies the method padarray uses to determine the values of the elements added as padding.

Value	Meaning
‘circular’	Pad with circular repetition of elements within the dimension.
‘replicate’	Pad by repeating border elements of array.
‘symmetric’	Pad array with mirror reflections of itself.

Example 1
Add three elements of padding to the beginning of a vector. The padding elements, indicated by the gray shading, contain mirror copies of the array elements.
将三个填充元素添加到向量的开头。由灰色底纹指示的填充元素包含数组元素的镜像副本。

a = [ 1 2 3 4 ];
b = padarray(a,[0 3],'symmetric','pre')

结果：

b ==
3 2 1 1 2 3 4
% 前3个数字3 2 1为填充的元素

OpenCV 实现matlab的padarray（A, padsize, ‘symmetric’）函数:
在C++代码中函数为：

//Replicate the boundaries of the InImage image
//复制输入图像的边界
//symmetric:Pad array with mirror reflections of itself.
Mat PadArray(Mat Image, int RowPad, int ColPad)
{   
    int n = Image.rows;
    int m = Image.cols;
    Mat temp1 = Mat::zeros(n, m + ColPad * 2, Image.type());
    Mat temp2 = Mat::zeros(n + RowPad * 2, m + ColPad * 2, Image.type());
    for (int i = 0; i < ColPad; i++)
    {
        Image.col(i).copyTo(temp1.col(ColPad - 1 - i));
        Image.col(m - 1 - i).copyTo(temp1.col(m + ColPad + i));
    }
    Image.copyTo(temp1.colRange(ColPad, m + ColPad));
    for (int j = 0; j < RowPad; j++)
    {
        temp1.row(j).copyTo(temp2.row(RowPad - 1 - j));
        temp1.row(n - 1 - j).copyTo(temp2.row(n + RowPad + j));
    }
    temp1.copyTo(temp2.rowRange(RowPad, n + RowPad));
    return temp2;
}

main函数：

int main()
{
    Mat C = (Mat_<double>(3, 3) << 1, 2, 3, 4, 5, 6，7，8，9);
    cout << C << endl << endl << endl;
    Mat B = PadArray(C, 1, 2);
    cout << B << endl;

    return 0;
}

运行结果：

在原有矩阵的周围进行镜像填充后的结果

padarray实现结果

其他说明：
关于padarray方法的其他重载函数以后再上传，这只是一个简单的实现。