Language
Country/Area
No result found
Block Matching Algorithm Revealed: Why is Video Compression So Efficient?
In the field of digital video processing, Block Matching Algorithm (BMA) plays a crucial role. This algorithm is mainly used to locate matching macroblocks in video frame sequences, and its core purpose is motion estimation. Motion estimation assumes that objects and backgrounds in video frames move over time, thereby generating corresponding objects in subsequent frames.
This process can reveal temporal redundancy in the video sequence, making inter-frame video compression more efficient.
When implementing the block matching algorithm, the current frame is divided into multiple macroblocks, and each macroblock is compared with the corresponding block in the previous frame and its adjacent blocks. In this way, a motion vector is generated that represents the movement of a macroblock from one position to another. The motion summary for all macroblocks is the motion estimate of the frame.
In the video compression process, it is crucial to choose an appropriate search range, which is determined by the "search parameter" p. Specifically, p represents the number of pixels surrounding the corresponding macroblock in the previous frame. If the value of p is larger, it means that there may be larger displacements, making it easier to find a good match, but this may also lead to increased computational complexity. Generally, the macroblock size is 16 pixels and the p-value of the search area is set to 7 pixels.
Motivation
During the motion estimation process, the calculation of action vectors can describe the transition from one 2D image to another on the one hand, and on the other hand, it can also predict image changes through motion compensation. This technology is the cornerstone of video compression standards such as MPEG 1, 2 and 4.
Video compression can effectively reduce the number of bits required for data transmission through motion estimation, since it is more efficient to transmit encoded difference images than fully encoded frames.
However, in the entire compression process, motion estimation is the most computationally expensive operation, so seeking a fast and computationally simple motion estimation algorithm has become an important requirement for video compression technology.
Evaluation indicators
The most commonly used cost functions when comparing one macroblock to another are Mean Absolute Difference (MAD) and Mean Squared Error (MSE). These indicators can help the algorithm evaluate the quality of the matching:
MAD = 1/N^2 * ∑(i=0 to n-1) ∑(j=0 to n-1) |C(i,j) - R(i,j)|
MSE = 1/N^2 * ∑(i=0 to n-1) ∑(j=0 to n-1) (C(i,j) - R(i,j))^2
Among them, N represents the size of the macroblock, and C(i,j) and R(i,j) are the pixels in the current macroblock and the reference macroblock respectively.
Algorithm
Since the 1980s, research on block matching algorithms has made significant progress, and a variety of efficient algorithms have emerged. The following is an introduction to several common algorithms:
1. Exhaustive Search
This algorithm calculates the cost function of each position in the search window and can find the most matching macroblock in the reference frame. However, its computational overhead is huge and it is the most cumbersome of all block matching algorithms.
2. Optimized Hierarchical Block Matching (OHBM)
This algorithm is based on the optimized image pyramid to speed up the full search process and improve efficiency.
3. Three Step Search (TSS)
TSS is one of the early fast block matching algorithms that significantly reduces the number of macroblocks that need to be evaluated by searching in multiple locations.
4. Four Step Search (FSS)
Compared with TSS, FSS performs better in terms of computational cost and peak signal-to-noise ratio (PSNR), and also uses a center-biased search method.
"With the advancement of video technology, how will the development of block matching algorithms affect the evolution of future video compression technology?"
The efficiency and quality of video compression rely heavily on the accuracy and computational efficiency of motion estimation, which in turn are closely related to the selected block matching algorithm. Therefore, understanding the pros and cons of various algorithms will have an important impact on the development of future video technology. How will the block matching algorithm affect future changes in digital video technology?
