Many of affine motion compensation techniques proposed thus far employ least-square-based techniques in estimating affine parameters, which requires a hardware structure different from conventional block-matching-based one. This paper proposes a new affine motion estimation/compensation framework friendly to block-matching-based parameter estimation, and applies it to an HEVC encoder to demonstrate its coding efficiency and computation cost. To avoid a nest of search loops, a new affine motion model is first introduced by decomposing the conventional 4-parameter affine model into two 3-parameter ones. Then, a block-matching-based fast parameter estimation technique is proposed for the models. The experimental results given in this paper show that our approach is advantageous over conventional techniques.

In this paper, we propose a method for removing block noise which appears in JPEG (Joint Photographic Experts Group) encoded images. We iteratively perform the 3D wiener filtering and correction of the coefficients. In the wiener filtering, we perform the block matching for each patch in order to get the patches which have high similarities to the reference patch. After wiener filtering, the collected patches are returned to the places where they were and aggregated. We compare the performance of the proposed method to some conventional methods, and show that the proposed method has an excellent performance.

The large data-transfer time among different cores is a big problem in heterogeneous multi-core processors. This paper presents a method to accelerate the data transfers exploiting data-transfer-units together with complex memory allocation. We used block matching, which is very common in image processing, to evaluate our technique. The proposed method reduces the data-transfer time by more than 42% compared to the earlier works that use CPU-based data transfers. Moreover, the total processing time is only 15 ms for a VGA image with 1616 pixel blocks.

Robust object tracking is required by many vision applications, and it will be useful for the motion analysis of moving object if we can not only track the object, but also make clear the corresponding relation of each part between consecutive frames. For this purpose, we propose a new method for moving object extraction and tracking based on the exclusive block matching. We build a cost matrix consisting of the similarities between the current frame's and the previous frame's blocks and obtain the corresponding relation by solving one-to-one matching as linear assignment problem. In addition, we can track the trajectory of occluded blocks by dealing with multi-frames simultaneously.

In this letter, a novel technique for fast block matching using a new matching criterion is proposed. The matching speed and image quality are controlled by the one control parameter called matching region ratio. An efficient matching scheme with a gradual voting strategy is also proposed. This scheme can greatly boost the matching speed. The proposed technique is fast and applicable even in the presence of speckle noise or partial occlusion.

In this paper, we propose block matching and learning for color image classification. In our method, training images are partitioned into small blocks. Given a test image, it is also partitioned into small blocks, and mean-blocks corresponding to each test block are calculated with neighbor training blocks. Our method classifies a test image into the class that has the shortest total sum of distances between mean blocks and test ones. We also propose a learning method for reducing memory requirement. Experimental results show that our classification outperforms other classifiers such as support vector machine with bag of keypoints.

The block based motion estimation technique is adopted by various video coding standards to reduce the temporal redundancy in video sequences. The core of that technique is the search algorithm implemented to find the location of the best matched block. Indeed, the full search algorithm is the most straightforward and optimal but computationally demanding search algorithm. Consequently, many fast and suboptimal search algorithms have been proposed. Reduction of the number of location being searched is the approach used to decrease the computational load of full search. In this paper, hybridization between an adaptive search algorithm and the full search algorithm is proposed. The adaptive search algorithm benefits from the correlation within spatial and temporal adjacent blocks. At the same time, a feature domain based matching criteria is used to reduce the complexity resulting from applying the pixel based conventional criteria. It is shown that the proposed algorithm produces good quality performance and requires less computational time compared with popular block matching algorithms.

An improved algorithm for fast motion estimation based on the block matching algorithm (BMA) is presented for use in a block-based video coding system. To achieve enhanced motion estimation performance, we propose an adaptive search pattern length for each iteration for the current macro block (MB). In addition, search points that must be checked are determined by means of directional information from the error surface, thus reducing intermediate searches. The proposed algorithm is tested with several sequences and excellent performance is verified.

This paper presents a personal identification method based on BMME and LDA for images acquired at anterior and posterior occlusion expression of teeth. The method consists of teeth region extraction, BMME, and pattern recognition for the images acquired at the anterior and posterior occlusion state of teeth. Two occlusions can provide consistent teeth appearance in images and BMME can reduce matching error in pattern recognition. Using teeth images can be beneficial in recognition because teeth, rigid objects, cannot be deformed at the moment of image acquisition. In the experiments, the algorithm was successful in teeth recognition for personal identification for 20 people, which encouraged our method to be able to contribute to multi-modal authentication systems.

In this paper, we propose new low-hardware-cost motion estimation with a large search range for VLSI multimedia processors. It reduces the hardware amount required for pixel comparison by reducing both the spatial-resolution and bit-resolution of pixel values. Low-hardware-cost block-matching criterion is also employed. To avoid performance degradation from low resolution, we introduce an "outlier" pixel with large overload quantization error in the search window, and a search position is excluded from the motion estimation if its corresponding search window block contains one or more outliers. The proposed motion estimation is easy to implement in VLSI multimedia processors, and it significantly reduces the hardware amount when the search range is larger than 6464. In MPEG2 MP@ML video compression with 128128 search range, it reduces the hardware cost to 1/144 that of the full search algorithm, while its degradation of peak signal-to-noise ratio is 0.32 dB.

The recent sight system requires high stabilization functions for the longer range of observation and the higher kill probability. To this end, it is necessary to compensate rotational disturbances which are not stabilized with the conventional 2-axes stabilization system. This paper proposes a simple method on the rotational motion estimation for the stabilization of the sight system.

In this paper, the false-peaks problem of the conventional correlation-based video tracking is investigated using a simple mathematical analysis. To reduce the false detection problem, a selective-attention correlation measure is proposed. The problem with the conventional correlation measures is that all pixels in the reference block image are equally treated in the computation of the correlation measures irrespective of target or background pixels. Therefore, the more the reference block image includes background pixels, the higher probability of false-peaks is introduced due to the correlation between the background pixels of the reference block and those of the input search image. The proposed selective-attention correlation measure has different consideration according to target and background pixels in the matching process, which conform with the selective-attention property of human visual system. Various computer simulations validated these analyses and confirmed that the proposed selective-attention measure is effective to reduce considerably the probability of the false-peaks.

This paper presents a technique for high-accuracy correspondence search between two images using Phase-Only Correlation (POC) and its performance evaluation in a 3D measurement application. The proposed technique employs (i) a coarse-to-fine strategy using image pyramids for correspondence search and (ii) a sub-pixel window alignment technique for finding a pair of corresponding points with sub-pixel displacement accuracy. Experimental evaluation shows that the proposed method makes possible to estimate the displacement between corresponding points with approximately 0.05-pixel accuracy when using 1111-pixel matching windows. This paper also describes an application of the proposed technique to passive 3D measurement system.

This paper presents a block-based motion vector search algorithm for video coding based on an interpolation scheme of search blocks. The basic idea of motion vector estimation between frames is to select a block in the previous frame that best matches a block in the current frame by minimizing the difference between them. In most of the search algorithms, however, the best-match block can only be on a pre-defined grid pattern. Although using a pre-defined pattern increases the search efficiency, it may also reduce the search accuracy. To balance the two aspects and to fully utilize the block information, we propose a strategy, which, instead of selecting from pre-defined blocks, searches for a best match interpolated from the pre-defined blocks. Experiment results demonstrate a better accuracy and efficiency of this search method than some commonly-used methods for different kinds of motion.

In this paper, we propose a fast full-search block matching algorithm for motion estimation, based on binary edge information. The binary edge information allows a faster search by reducing the computational complexity. It also reduces error, which is generated by the block located on the boundary of moving objects. After we transform the input image into an edge-based image using Sobel masks, we convert the result into a binary edge image using median-cut quantization. We then perform block matching using the binary edge image. If there exists blocks such that the error of the binary block matching exceeds threshold, we only perform edge intensity-based block matching within those blocks. We improve computational efficiency by eliminating an unnecessary searching process in no-motion regions. Simulation results have shown that the proposed method reduces the computational complexity and provides similar PSNR performance to the Full Search Block Matching Algorithm (FS-BMA)

In this letter we propose a new fast matching algorithm that has no degradation of predicted images such as found in the conventional full search (FS) algorithm, so as to reduce the amount of computation of the FS algorithm for motion estimation in real-time video coding applications. That is, our proposing algorithm reduces only unnecessary computations in the process of motion estimation without decreasing the prediction quality compared to the conventional FS algorithm. The computational reduction comes from rapid elimination of impossible motion vectors. In comparison to the FS algorithm, we obtained faster elimination of inappropriate candidate motion vectors using efficient matching units based on image complexity. Experimentally, we demonstrated that the unnecessary computations were removed by about 30% as compared to the other fast FS algorithms.

An advanced center biased search algorithm for block motion estimation is proposed in this letter. It adopts an innovative center biased search strategy to get correct motion vector. The computational complexity is reduced by strict application of the unimodal error surface assumption and half stop technique. Experimental results show that proposed algorithm has improved performance as compared to the conventional block matching algorithms.

We propose a new and fast full search (FS) motion estimation algorithm for video coding. The computational reduction comes from sequential rejection of impossible candidates with derived formula and subblock norms. Our algorithm reduces more the computations than the recent fast full search (FS) motion estimation algorithms.

In this paper, we propose a multi-level block matching algorithm using motion information in blocks. In the proposed algorithm, the block-level is decided by the motion degree in the block before motion searching procedure, and then adequate motion searching performs according to the block-level. Which improves computational efficiency by eliminating an unnecessary searching process in no motion or low motion regions, and brings more accurate estimation results by deepening motion searching process in high motion regions. Simulation results show that the proposed algorithm brings the lower estimation error--about 20% MSE reduction--with the fewer blocks per frame and the lower computational loading--about 98% operational amount reduction--than full search block matching algorithm with constant block size.

To reduce an amount of computation of full search algorithm for fast motion estimation, we propose a new and fast matching algorithm without any degradation of predicted images. The computational reduction without any degradation comes from adaptive matching scan algorithm according to the image complexity of the reference block in current frame. Experimentally, we significantly reduce the computational load compared with conventional full search algorithm.