DOI: 10.4236/jcc.2018.612012 120 Journal of Computer and Communications large. All the image data has to be loaded into memory for calculation, so it is not only computationally large, but also prone to memory leaks. In view of this, this paper proposes a density peak clustering algorithm which combines the characteristics of HSV histogram, uses the HSV histogram to simplify calcula- tion and effectively improves the quality and efficiency of keyframeextraction.
In this paper we discussed several ways of hiding the secret data inside the cover medium such as image, audio and video. The proposed method for data hiding uses LSB techniques and statistical features for keyframeextraction and AES for encryption which results in more secure and robust technique for data hiding as average MSE and PSNR value have been reduced much as compared to HLSB and LSB techniques. We can conclude that the proposed system is more effective for secret communication over the network channel, as keyframeextraction method would not be known to hackers as it is completely new technique for video steganography, which is the main strength of my proposed method, so they will not be able to find data hidden frames easily. Hence it is clearly perceived from the result that the distortions occurred in the cover-video by applying the proposed algorithm is highly imperceptible to human eyes. Hence the proposed video steganography method is proved to be apparent to transfer highly confidential data like banking details, medical reports, military data and other important data.
The work area of keyframeextraction is so wide and rich technology. Many techniques for keyframe detection have been reported in so far. One of the possible approaches is the detection of video shots and the first frame of a shot is chosen as keyframe . Next interesting technique is the work of Seung et al. In this work, shot boundary was detected in low pass filtering in histogram space and keyframe was selected by using adaptive temporal sampling. . Furthermore, Mohamed and his colleagues published another histogram approach to extract keyframe . In his work, histograms of RGB channels of each consecutive frame were firstly estimated and found their differences. From the different values, threshold was calculated and compare with different value to choose keyframe. Later, Hong et al report appearance based keyframe detection in  in which keyframe is extracted by using pixel wise, global histogram, local histogram, feature matching and Bags of Words (BoW). Another wavelet approach for keyframe detection is issued by Gianluigi et al in . In their work, keyframe detection algorithm determined the complexity of video sequence in term of video content change. Three descriptor, colour histogram, wavelet and edge direction histogram were used to express the frame visual contents. In the last and one of the related works of our proposed method, the work of Khushboo et al  is wanted to present. This work is very sample but efficient. It is pixel wise processing in which the gradient differences of related pixels of consecutive frames is firstly estimated and then compare with threshold to determine keyframe.
In this section, the results of our scheme are compared with some of the popular non-visual attention-based tech- niques. For this purpose, the experiments are conducted based on 20 videos selected from the Open Video Project. The videos belong to different genres including histori- cal, education, and documentary. These videos were part of the data set used by many authors [12–14] in the eval- uation of their video summarization algorithms. All the videos are in mpeg format. The information about the data set videos is presented in Table 2. For the evaluation data set of Table 2, the proposed scheme has been com- pared with four other schemes for keyframeextraction: OV , DT , STIMO , and VSUMM .
Keyframes play an important role in video abstraction. Keyframes are a set of salient images extracted from video sequences. They provide a simple yet effective way of summarizing the content of videos for browsing and retrieval and are also widely used in video abstraction due to their compactness. Much research has been conducted in the past few years in understanding the problem of key- frameextraction and developing effective algorithms. Although simple and computationally efficient sampling- based methods may produce no keyframes for a shot, yet semantically producing too many keyframes with identical content to represent a long static segment thus failing to effectively represent the actual video content. The keyframe extraction algorithm discussed, here will focus only on techniques that take into account the underlying dynamics, to different degrees and from varying viewpoints, of the video sequence. The algorithm of keyframe extraction is described as follows:-
Based on the characteristics of video data and the characteristics of video data, video segmentation and keyframeextraction based on compressed video stream are adopted. Based on the video search technology of massive video information, according to the characteristics provided by the user quickly browse and play, it goes without saying that there is a very large and beautiful application prospects. However, in order to make the whole system into practical use, future research needs to address the following issues: compression techniques for compressing video streams require higher compression techniques that allow users to view clearer video streams; feature to enrich the content of the content space, which is an important part of future research.
Abstract. Human action recognition is an important part of intelligent video analysis. Recently, deep learning has made significant progress in this field, and state-of-the-art methods are based on the two-stream convolutional networks. In long-term action recognition, existing approaches mainly use video frames obtained by averaging or sampling as input, which may lose important information in the sampling interval. By defining the amount of video information, we propose a method of segment division and keyframeextraction for action recognition is proposed, where multi-temporal-scale two-stream networks are used to extract features. We achieve 94.2% accuracy on a widely used action recognition benchmark (Split1 of UCF-101).
Abstract: Keyframeextraction has been recognized as one of the important research issues in video information retrieval. Although progress has been made in keyframeextraction, the existing approaches are either computationally expensive or ineffective in capturing most important visual content. Video summarization aimed at reducing the amount of data that must be examined in order to retrieve the information desired from information in a video, is an essential task in video analysis and indexing applications. We propose an innovative approach to the selection of representative (key) frames of a video sequence for video summarization In this paper, we discuss the importance of keyframe selection; and then briefly review and evaluate the existing approaches. To overcome the shortcomings of the existing approaches, we introduce a new algorithm for keyframeextraction.
The accurate segmentation of shots in a video sequence is fundamental and an essential functionality for numerous video retrieval and management tasks . Many researchers have proposed algorithms to perform shot boundary detection based on certain features extracted from video frames, such as pixel differences, edge differences, color histograms, etc. Moreover, comparative surveys have also From a learning theory perspective, it is a natural approach to combine such promising features in order to decide whether a boundary exists or not within a given video sequence. Researchers have actively developed different approaches for intelligent video management, including shot transition detection, keyframeextraction, video retrieval, etc.
original data to aid the selection of the key-frames that can be used to reconstruct the original animation with smaller error. Usually, an animation sequence is charac- terised by a large amount of information. For computa- tional efficiency, the animation sequence is projected to a lower-dimensional space where all frames of the sequence are represented as points of curves defined in the new lower-dimensional space. Then, the curves in the lower- dimensional space are sampled and these sampled points are used to compute the Gaussian curvature values. Next, the points with the largest curvature value are selected as candidate key-frames. Finally, a key-frame refinement method is employed to minimise an error function which incorporates visual saliency information. The aim of a visual saliency is to identify the regions of an image which attract higher human visual attention. Lee Huang et al.  expanded this idea to 3D video and computed mesh saliency for use in a mesh simplification algorithm that preserves much information of the original input. More recently, visual saliency has also been used in 3D key- frameextraction, in the method proposed by Ferreira et al. in .
Tianming Liu et & al. presented that The keyframe is a simple yet effective form of summarizing a long video sequence. The number of key frames used to abstract a shot should be compliant to visual content complexity within the shot and the placement of key frames should represent most salient visual content. Motion is the more salient feature in presenting actions or events in video and, thus, should be the feature to determine key frames. In this paper, we propose a triangle model of perceived motion energy (PME) to model motion patterns in video and a scheme to extract key frames based on this model. The frames at the turning point of the motion acceleration and motion deceleration are selected as key frames. The key-frame selection process is threshold free and fast and the extracted key frames are representative .In this paper, they have presented a novel key-frame-extraction approach that combines motion-based temporal segmentation and color-based shot detection. The turning point of motion acceleration and deceleration of each motion pattern is selected as a keyframe. If a shot is static, the first frame of the video shot is selected as a keyframe. With this approach, both the number of key frames and the location of the key frames in a given video are determined automatically by the perceived motion patterns of the video. The proposed approach is threshold free and also fast since motion information in MPEG video can be directly utilized in the motion analysis. Our future work to improve the proposed algorithm includes the integration of color-change analysis and audio cues.
Abstract— This paper presents a new approach for text-based video content retrieval system. The proposed scheme consists of three main processes that are keyframeextraction, text localization and keyword matching. For the key-frameextraction, we proposed a Maximally Stable Extremal Region (MSER) based feature which is oriented to segment shots of the video with different text contents. In text localization process, in order to form the text lines, the MSERs in each keyframe are clustered based on their similarity in position, size, color, and stroke width. Then, Tesseract OCR engine is used for recognizing the text regions. In this work, to improve the recognition results, we input four images obtained from different pre-processing methods to Tesseract engine. Finally, the target keyword for querying is matched with OCR results based on an approximate string search scheme. The experiment shows that, by using the MSER feature, the videos can be segmented by using efficient number of shots and provide the better precision and recall in comparison with a sum of absolute difference and edge based method.
object based video segmentation. Due to being a different sementic levels, keyframeextraction and object segmentation are generally implemented separately and independently in a conventional method. This conventional method hence overlooked the inherent relationship between key frames and the objects. In the proposed approach, small number of key frames are extracted within a video shot for maximizing the divergence between the video object in a feature space. Which result in robust and efficient object segmentation. This method hence utilize the temporal based as well as the object6 based video segmentation which is very helpful for content based video analysis and structured video representation. Theoretical analysis and the experimental results on the standard test video reveals the effectiveness of the proposed method.
Key frames provide the effective summary information for video retrieval and browsing, which mostly represents the main content for video sequences. The statistical results prove that color, as a feature that represents image content, proved widely used in the keyframeextraction. Stereo correspondence has been an indispensable research direction in the field of computational stereo vision. Stereo vision is introduced for the purpose of improving previous keyframeextraction algorithms, which is the key contribution of our work presented in this paper. A novel image similarity measurement is constructed by combining the color moment with SAD. Experimental results demonstrate that the average evaluation coefficient obtained by the improved algorithm in this paper reach up to 95% in the situation of camera stillness and motion. The novel algorithm solves the issue of redundancy which stems from the large changes of content among consecutive frames in the process of extracting key frames using monocular cameras, and performs more robust compared with traditional algorithms.
In this paper, we summarize the shortcomings of several keyframeextraction algorithms, and propose a hierarchical filtering algorithm for video image keyframe, which divides the keyframe into shallow screening and deep screening. Shallow screening factors are image clarity, the speed of the aircraft and the number of camera frames. Shallow screening of the keyframe, and then according to the keyframe overlap, the width of the baseline and other factors that can affect the three- dimensional reconstruction of the effect of deep screening.
The future scope of this approach is very high and the improvements can be done in future. Since the method is technical based, the output performance depends purely on the developers based on their selection of distance (or difference) measure. In the future, new distance measures may be combined with appropriate frame descriptors to get a better output than the present one. The output success rate is determined by evaluating on how efficiently the extracted key frames are able to produce the detailed content of the input video shot when all of them are combined to form the video. While at the same time, emphasis is to be laid upon the minimizing the number of key frames such that the summarized output video is of small size. The audio is not embedded and only the visual details are considered in the current project. In the future, the developers may take the project to a new level by incorporating the audio feature as well.The size of any video shot can he reduced to a significant level by the keyframeextraction feature, but it is a loss full compression technique. The summarized video is able to highlight the key contents of the original video shot. The number of key frames extracted is considerably less, as compared to the sequential frames that were obtained initially which led to the compression of the final video. Depending on the input video frame, variable numbers of key frames are extracted. The most significant observation is that the work does not take into consideration the audio pan that is available in the input video shot.
In the current era, most of the digital information in the form of multimedia with a giant share of videos. Videos do have audio and visual content where the visual content has number of frames put in a sequence. Most of the consecutive frames do have very little discriminative contents. In video summarization process, several frames containing similar information are needed to get processed. This leads to redundant slow processing speed and complexity, time consumption. Video summarization using key frames can ease the speedup of video processing. In this paper, novel keyframeextraction method is proposed with Linde-Buzo-Gray (LBG) codebook generation techniques of vector quantization with ten different codebook sizes. Experimentation done with the help of the test bed of videos has shown that higher codebook sizes of LBG have given better completeness in keyframeextraction for video summarization. Experimental results are also discussed to represent the validity of the proposed method for video content summarization.
First of all, we demonstrate the performance of keyframeextraction video captured on the particular timing of entering the students, staff, and research scholars in the laboratory. Mostly the entities are entering and leaving constantly with making groups with different person with different time. That form of group is to be discovered as a frequent group. While generating frames from the video several redundant frames are also generated with no information or lesser information that should be omitted or discarded
After shot segmentation, the keyframe representing the particular shot needs to be extracted. The so-called keyframe extracted from the original video frames in the sequence can reflect a synopsis of the shot. Because the keyframe plays an important role in video post-production, the keyframeextraction technology is a hot research topic in multimedia information technology field. From the perspective of information theory, video frames with difference between each other carry more information than the video frames similar to each other, so the keyframeextraction is considered more about the dissimilarity between video frames. According to the complexity of the shot content, one or more key frames from a shot can be extracted, and used to summarize the content. There are many methods to extract keyframe. The first frame, the last frame or the intermediate frame can be regarded as the keyframe if there is little change in a shot content. But in the case of large changes in content, extraction technology usually based on motion analysis, image information, clustering, shot activities analysis method should be accepted [8-10].
Ganesh. I. Rathod, Dipali. A. Nikam , proposed An Algorithm for Shot Boundary Detection and KeyFrameExtraction Using Histogram Difference. This paper work a Square histogram based model is developed using frame segmentation and automatic threshold calculation. In this paper the keyframe is extracted by using a reference frame approach per shot. A total of around 40 videos of different types are tested on this model and the model is able to detect all shot boundaries and is storing the suitable frames as keyframes to represent the video summary. An efficiency of almost 95% to 98% is observed using this algorithm.