比例图像商图像引文

提取图像细节的两种方法 - 天律界中子之不写博客不读书 - CSDN博客 https://blog.csdn.net/u012494876/article/details/80588823
《Digital Photography with Flash and No-Flash Image Pairs》乘性细节的应用例子;
1.A flash image captures the high-frequency texture and detail, but changes the overall scene appearance to cold and gray. The no-flash image captures the overall appearance of the warm candlelight,but is very noisy. We use the detail information from the flash image to both reduce noise in the no-flash image and sharpen its detail
2.

image.png
To transfer this detail we begin by computing a detail layer from the flash image as the following ratio: 𝐵𝑎𝑠𝑒 is computed using the basic bilateral filter on 𝐹. The ratio is computed on each RGB channel separately and is independent of the signal magnitude and surface reflectance. The ratio captures the local detail variation in 𝐹 and is commonly called a quotient image [Shashua and Riklin-Raviv 2001] or ratio image [Liu et al. 2001] in computer vision. Figure 5 shows that the advantage of using the bilateral filter to compute 𝐹 𝐵𝑎𝑠𝑒 rather than a classic low-pass Gaussian filter is that we reduce haloing.

《The Quotient Image: Class-Based Re-Rendering and Recognition with
Varying Illuminations》
1.An ideal class is a collection of 3D objects that have the same shape but differ in the surface albedo function.

2. Clearly, Q is illumination invariant. In the absence of any direct access to
   the albedo functions, we show that Q can nevertheless be recovered, analytically, given a bootstrap set of images. Once Q is recovered, the entire image space (under varying lighting conditions) of object a can be generated by Q and three images of object b. The details are below.

   
   image.png
3. we wish to create the color image space of that object under varying illumination. To that end, we will make the assumption that varying
   illumination does not affect the saturation and hue composition of the image, only the gray-value distribution (shades of color) of the image.