I am researching on RGB-D images processing. These images are commonly gathered by Kinect cameras and include a rgb image and its corresponding depth map. I am interested to know is there any relation between color and depth of these points? The relation could be model based or casual.
There is no necessary relationship. One can structure a situation where there is one, but it is not the default situation.
In my experience the depth of each pixel can be used to obtain specific data, but the most important information used form image processing and video signal processing are the gray value of pixels in RGB or other color space.
Computer graphics has "alpha blending" i.e RGBa to provide depth perception which I feel is used by artist also. If you open a bmp file with 32 bit data per pixel , 24 bits are for RGB( 8 bit each)and 8 more is for alpha.
http://en.wikipedia.org/wiki/Alpha_compositing
Do cameras write this data (on to the buffer) ? I am not sure ... If it is written as a part of digitization (A-D conversion) then it may be causal !!
Your answer is more about display, not image analysis. Yes, one can give an illusion of depth using alpha, but I know of no device that uses this. Color/grey conveys true depth only if that has been established a priori; structured light or conversion of depth to color by the acquisition device. Rangefinders often convert depth to grey for instance.
No in general there is no relation. If the image is a disparity map, then brighter part means low depth and vice versa. If it is a normal gray image, you can try to find depth using shape from shading and intrinsic image techniques...
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing