White Balance Principle(4)
In the previous article, the general principles of the white balance algorithm are described. The three algorithms are relatively simple, and the amount of computation is not large, but each has its own advantages and disadvantages, and further more efficient hybrid algorithms are generated, such as: luminance-weighted gray-scale world algorithm and total reflection algorithm. Quadratic Combining Luminance Weighted Gray World & Prefect Reflector Assumption. Look at such a long name, the algorithm is very complicated, the white balance correction effect is quite good, and it is convergent, it will not bring too much loss when processing the image, but unfortunately the amount of calculation is huge, The requirements for hardware resources are too high.
The algorithm is quite boring. If the non-technical engineering is not interested in mathematics, you can skip it directly. You can generally know that this is the case. But the conclusion I can give is that the better the final algorithm, the higher the complexity, the larger the computation, and the higher the hardware requirements. The specific implementation also needs to balance the three aspects of white balance correction ability, algorithm execution efficiency, and processor hardware performance.
Through the above understanding, you will find that if the ISP (Image Signal Processor) is capable of high performance, the space used by the white balance algorithm will be much larger. In some cases, the white balance is not accurate. To a certain extent, it is really related to the performance of the ISP. Of course, here is also to look at the skills of various vendors in algorithm optimization. Generally speaking, the flagship processing performance difference of each brand in the same era will not be particularly large. Although the software is not easy to see, the importance is undoubted. Whether the overall performance of the hardware can be fully utilized is the key.
For example, in the era when the SLR was just digitalized, the camera's image processor performance was relatively low, and it was difficult to withstand the high-strength white balance algorithm, so many SLR cameras (such as Canon 1D, Nikon D2, Olympus) E-1, etc.) There is a white balance sensing device on the fuselage (that is, the small white point on the front of the fuselage), which can help improve the white balance accuracy. Later, as the performance of the camera image processor soared, probably from the Expeed generation of Fujitsu's Expeed generation processor, the external white balance sensor was removed. Through more and more RGB metering partitions, with more powerful processors for more accurate color temperature correction. Here are two more words. The more partitions, the more accurate the white balance sampling process will be, but it will also bring about a surge in computational complexity, from the initial few partitions to the 91,000-pixel RGB sensor used on the D800. At the same time, the metering and white balance calculations are completed at the same time, and even the power can be used for face recognition. The latter part of the ARM architecture Expeed 3 is the biggest hero.
For example, DC, mobile camera, such continuous viewfinder camera, is applied to the latter image with the processing result of the previous frame image. The implementation is not the same as the single photometric sensor on the SLR. . This is caused by structural differences in the product itself.