I read thru the patent as well.
It pertains to the need to read data from a large mp sensor without overwhelming the buffer memory with huge quantities of data. Typical thinning methods cause a color stripe, this one is designed to avoid that.
This allows for more images to be captured before a buffer fills, obviously a issue for a large mp camera.
Expect this to be used in a high mp camera, its basically firmware.
The patent number is actually 2013-012850, Igami dropped the leading zero.
Here is a quote from the translated text.
[Problem to be solved by the invention]
It is necessary to memorize a picture signal by four plains of R/G1/G2/B, and to memorize the picture signal of each plain for every frequency band further in the technology of a Patent document 2. Therefore, a mass memory is needed and it is not preferable in respect of cost. Since there are also many required amounts of memory access, when rapidity, such as continuous shooting, is the processings called for, there is a possibility of reaching the maximum of a memory access zone and causing deterioration of processing speed.
In order to ease this problem, it is possible to, apply the thinning technology of a chrominance signal as shown in a Patent document 1 to the composition of a Patent document 2 for example, and to reduce required memory space and the amount of memory access.
However, in order to interpolate the thinned-out chrominance signal in the technology of a Patent document 1 from the chrominance signal of an adjacent pixel and to restore, a difference arises in the frequency band of a chrominance signal, and the frequency band of a luminance signal, and there is a problem that the color stripe not existing will occur in an object.