最近写这块代码的性能优化。
public static Mat array2Mat(double[][][] data) {
int height = data.length;
int width = data[0].length;
Mat mat = new Mat(height, width, CvType.CV_8UC3);
double[] mat_data = new double[width * height * 3];
for (int i = 0; i < height; i++) {
for (int j = 0; j < width; j++) {
mat_data[i * width * 3 + j * 3] = data[i][j][0];
mat_data[i * width * 3 + j * 3 + 1] = data[i][j][1];
mat_data[i * width * 3 + j * 3 + 2] = data[i][j][2];
}
}
mat.put(0, 0, mat_data);
return mat;
/*int height = data.length;
int width = data[0].length;
Mat mat = new Mat(height, width, CvType.CV_8UC3);
for (int i = 0; i < height; i++) {
for (int j = 0; j < width; j++) {
mat.put(i, j, data[i][j]);
}
}
return mat;*/
}
上面代码明显比下面代码快,不止一倍。
同样,get函数也是如此;
public static double[][][] matGray2Array(Mat mat) {
int width = mat.cols();
int height = mat.rows();
double[][][] data = new double[height][width][1];
byte[] grayData = new byte[height * width];
mat.get(0, 0, grayData);
for (int i = 0; i < height; i++) {
for (int j = 0; j < width; j++) {
data[i][j][0] = grayData[(i * width + j)];
data[i][j][0] = data[i][j][0] >= 0 ? data[i][j][0] : data[i][j][0] + 256;
// data[i][j] = mat.get(i, j);
}
}
return data;
}
注:byte的范围是-128到127
,取的时候是按照byte取,放的时候按照double放;
调试发现,-128到-1
的Byte数据转换到double数据,需要映射到128到255
(double取的时候整体范围是0到255
)。