测试图像到特征脸空间的投影究竟意味着什么？(What does the projection of test image into the eigenface space, exactly mean?

测试图像到特征脸空间的投影究竟意味着什么？(What does the projection of test image into the eigenface space, exactly mean?)

在使用特征脸技术的人脸识别中，测试图像（未知人脸图像）投影到特征脸空间的准确性是什么意思？

In the face recognition using eigenfaces technique, what does the projection of test image (unknown face image) into the eigenface space, exactly mean?

最满意答案

这意味着将测试图像表示为特征脸的加权和。

要计算投影：将测试图像乘以每个特征脸，逐像素，然后求和结果像素以获得该特征脸的系数。 “投影到特征脸空间”是这些系数的列表。

为了从它的特征脸空间投影中重建测试图像，将每个特征脸乘以它的系数，然后将所有得到的图像加在一起。

请注意，上述过程假定您的特征脸矢量已经标准化为单位长度。

It means representing test image as a weighted sum of eigenfaces.

To calculate the projection: multiply the test image by each eigenface, pixel-by-pixel, and sum the result pixels to get the coefficient for that eigenface. The "projection onto eigenface space" is the list of these coefficients.

To reconstitute the test image from its eigenface space projection, multiply each eigenface by its coefficient and then just add all the resulting images together.

Note that the above procedures assume that your eigenface vectors have been normalized to unit length.