Fast Bit-Plane Matching And Matching Pictures With Different Exposures In High Dynamic Video

Due to the limitation of the sensor, traditional digital video camera can only capture low dynamic videos. That will cause some details of the picture lost. With the developing of the technology, people care more about the video quality, and high dynamic video camera is proposed. The most difficult thing in generating high dynamic pictures is to align pictures with different exposures. There is no solution now can register rotational motion with a low operation amount. Motion estimation is a very important algorithm in digital video camera, because it has a big effect on video compression and image stabilization. Bitwise operation needs less resource and is very efficient in operation. It improves the processing efficiency. So it is very necessary to use bitwise operation in motion estimation.This paper first focus on the problem of the matching precision is low to only use one bit-plane. And we test the matching result with different bit planes base on grey code. Based on the fact that the diamond search can search very fast and most misalignments with grey code bit-plane matching can be corrected. We propose a fast mixed bit-plane matching method. The simulation result shows that our method has the best matching accuracy with other bit-plane matching methods. The operation amount of our method is little higher than the one bit-plane matching method and lower than others.We propose a double matched method to match picture with different exposures. Our method first aligns pictures with same exposure and use affine transformation to compensate the motion. Then we use Median Threshold bitmap to do matching with different exposure pictures. The simulation result shows that compare to MTB, our method have a better performance to solve the rotational motion. Our method’s operation time is very close to MTB, far below the time of using speeded up robust features.