Shaifali Madan Arora

Real time human face detection and tracking

This paper describes the technique for real time human face detection and tracking using a modified version of the algorithm suggested by Paul viola and Michael Jones. The paper starts with the introduction to human face detection and tracking, followed by apprehension of the Vila Jones algorithm and then discussing about the implementation in real video applications. Viola jones algorithm was based on object detection by extracting some specific features from the image. We used the same approach for real time human face detection and tracking. Simulation results of this developed algorithm shows the Real time human face detection and tracking supporting up to 50 human faces. This algorithm computes data and produce results in just a mere fraction of seconds.

Comparative analysis of motion estimation algorithms on slow, medium and fast video sequences

Motion estimation is the key component in the entire process of video compression. The growing need of compression has led to the development of various fast block based motion estimation algorithms. Based on the motion activity in the consecutive frames video sequences could be broadly divided in to three categories- slow, medium and fast video sequences. The current work gives a comparative analysis of various motion estimation algorithms on all the three categories. The motion estimation algorithms which are analyzed are Full Search (FS), Three Step Search (TSS), New Three Step Search (NTSS), Diamond Search (DS) and Adaptive Rood Pattern Search (ARPS). The performance of these algorithms is considered in terms of Peak to Signal Ratio (PSNR), number of computations and the time required to find motion vectors per frame. All the algorithms are compared and also the suitability of various algorithms is established for each of the categories of slow, medium and fast video sequences. The conclusions are drawn for the overall optimization of performance based on the mentioned factors.

Survey of fast block motion estimation algorithms

Tremendous advancements in video capturing and display technologies and increased video applications in all arenas of life have raised the demand for enhancement in the field of video compression. Motion Estimation (ME) a key component in most of the video data processing based applications, has led to an ongoing research in this field. Lots of algorithms to estimate block based motion and criterias to find best matching block had been developed. The current work reviews the advantages, disadvantages and various issues pertaining to these algorithms. Also the factors that have their impact on accuracy and efficiency of motion estimation like block matching criteria, edge matching in blocks, correlation between the neighboring blocks, pixel sub-sampling, size of search window, size of blocks and zero motion pre-judgment have been discussed. Apart from all of these the applicability, advantages and disadvantages of various block matching criteria are reviewed and mentioned.

Dynamic Pattern Search Algorithm with Zero Motion Prejudgment for Fast Motion Estimation

In the development of fast block based motion estimation (BME) algorithms, the focus is always on reduction of computational burden with quality as good as that of Full Search algorithm. Various fixed search BME algorithms like TSS, DS etc. Have been proposed in the literature for fast motion estimation but these suffer from over or under search for slow or fast motion video sequences. For quick determination of motion vector of fast moving blocks large search patterns would be helpful but this may cause unnecessary searches for small motion. Therefore a Dynamic Pattern Search algorithm had been suggested that uses the spatial coherence of the left adjacent block and temporal coherence of the collocated block from the reference frame to adjust the search pattern size. Further it was found that large numbers of blocks especially in slow motion sequences are zero motion blocks. Their early determination enhances the speed of motion estimation. A dynamic zero threshold determination model is implemented in this paper to speed up the motion estimation in Dynamic Pattern Search Algorithm. Simulation results clearly indicates 70-95% speed gain for slow motion sequences.

A Novel Hybrid Approach for Fast Block Based Motion Estimation

The current work presents a novel hybrid approach for motion estimation of various video sequences with a purpose to speed up the entire process without affecting the accuracy. The method integrates the dynamic Zero motion pre-judgment (ZMP) technique with Initial search centers (ISC) along with half way search termination and Small diamond search pattern. Calculation of the initial search centers has been shifted after the process of zero motion pre-judgment unlike most the previous approaches so that the search centers for stationary blocks need not be identified. Proper identification of ISC dismisses the need to use any fast block matching algorithm (BMA) to find the motion vectors (MV), rather a fixed search pattern such as small diamond search pattern is sufficient to use. Half way search termination has also been incorporated into the algorithm which helps in deciding whether the predicted ISC is the actual MV or not which further reduced the number of computations. Simulation results of the complete hybrid approach have been compared to other standard methods in the field. The method presented in the manuscript ensures better video quality with fewer computations.

A new fast motion estimation algorithm using adaptive size diamond pattern search with early search termination

In this paper, a new dynamic zero motion prejudgment (ZMP) with adaptive diamond pattern search-based algorithm is suggested to enhance the search efficiency and accuracy of motion estimation (ME) in video coding. Firstly, a dynamic ZMP technique is proposed for early identification of the stationary blocks. For non-stationary blocks, a new initial search centre prediction technique is suggested. This new search centre has high probability to be near actual MV. Its distortion is compared against a dynamically predicted threshold to check if this location could be the position of actual MV. If so, search is terminated thereafter, otherwise a variable size diamond pattern is suggested to swiftly attain the global minima. Experimental results show 95% to 99% speed gain of proposed algorithm with only 0.007-0.7 dB PSNR and 0.0001-0.0073 SSIM degradation over full search. Also, the proposed algorithm shows very promising results over other fixed and dynamic search algorithms.

Dynamic Two Level Threshold Estimation for Zero Motion Prejudgment: A Step Towards Fast Motion Estimation

In most video sequences, especially containing slow motion, a large number of blocks are stationary. Early determination of these blocks may save large number of computations in any motion estimation (ME) algorithm. The decision for declaring a block to be stationary can be made by comparing the block distortion with a predetermined threshold whose large or small values may affect the speed and accuracy of a ME algorithm. Accurate prediction of this threshold proposes a challenging problem. In this manuscript, a dynamic two level threshold estimation technique has been proposed. This two level scheme not only detects constant variations in the neighboring blocks but is also capable of detecting stationary blocks with abrupt variations. Performance of the proposed technique is evaluated by implementing ZMP before ME process in adaptive rood pattern search (ARPS) algorithm. Simulation results show better performance of proposed technique in comparison to single level dynamic threshold predictor and fixed threshold predictor.