1. bookAHEAD OF PRINT
Dettagli della rivista
License
Formato
Rivista
eISSN
2444-8656
Prima pubblicazione
01 Jan 2016
Frequenza di pubblicazione
2 volte all'anno
Lingue
Inglese
Accesso libero

Retrieval and Characteristic Analysis of Multimedia Tester Based on Bragg Equation

Pubblicato online: 15 Jul 2022
Volume & Edizione: AHEAD OF PRINT
Pagine: -
Ricevuto: 10 Mar 2022
Accettato: 21 May 2022
Dettagli della rivista
License
Formato
Rivista
eISSN
2444-8656
Prima pubblicazione
01 Jan 2016
Frequenza di pubblicazione
2 volte all'anno
Lingue
Inglese
Introduction

There are two kinds of multimedia information retrieval technology: text-based information retrieval and content-based information retrieval. The former focuses on the semantic description of information content, and carries out information retrieval by establishing keywords or text titles and some additional description information; The latter focuses on the feature extraction of multimedia information content, directly analyzes the content of multimedia information, extracts features and semantics, and uses content features to establish an index for retrieval. At present, multimedia information retrieval generally separates media information and often analyzes only one kind of media information, which may lead to the loss of some multimedia information and affect the accuracy of retrieval. Multimedia is the first mock exam of multimedia, such as text, image, video and audio. Each mode expresses rich semantic information, and usually does not appear in isolated form. They usually exist in one information unit to reinforce each other to transmit information[1].

Multimedia information retrieval system refers to the retrieval of discrete media represented by text information and continuous media represented by images and sounds. In order to achieve a better retrieval effect, the following key technologies must be solved. Information modeling is the modeling of application information using computer-based symbolic structures. Multimedia information retrieval relies on the organization of multimedia information, and the quality of multimedia information organization determines its retrieval efficiency to a certain extent. Common multimedia objects are stereotyped composite objects, which themselves can use multiple data models. The main models are: hypertext model, document model and information meta model. When performing full-text retrieval and free-text query on text information, users only need to submit a query request to find all documents involving the keyword. The result of a free-text query is a list of documents sorted by sequence value, with the most relevant probability at the top. The sequence is calculated according to a probability formula based on “lexical similarity”. The image information can be searched by color, shape, texture and position in the image[2].

Multimedia synchronization technology is to solve the problem of how to display the spatial combination of multimedia. Especially in the system using the client/service mode, various media sources are distributed in different databases, and the multimedia synchronization technology is to properly combine the data from different databases according to the arrangement of time sequence and spatial buffer address.

With the development of multimedia information retrieval technology, people's ability of information processing is increasing. Text, image, video and audio are the main parts of multimedia information. How to fuse, analyze and retrieve multimedia information has become a research hotspot. This paper emphasizes the advantages of multimedia information fusion index and retrieval, and summarizes the related technologies. At present, it is possible to realize some comprehensive multimedia information retrieval systems, but there are still many challenges in technology, and the retrieval efficiency needs to be studied and improved. Generally speaking, the research of multimedia content-based analysis, processing and retrieval is still in the primary stage, and it still needs a hard and long process to achieve real practicality[3].

Because multimedia retrieval algorithms usually face the challenge of real-time processing, many researchers have studied how to optimize various multimedia retrieval algorithms. ZhangW N and others implemented a parallel version of surf algorithm on multi-core processors[4]. Bekhet S et al. Accelerated SIFT algorithm by using general graphics processor (GPGPU)[5]. Tripathy S K and others proposed two parallel algorithms of SIFT for multi-core system, and implemented the corresponding algorithms on 8-core CPU and 32-core simulator[6]. These algorithms are optimized for a special application scenario, but only for certain application requirements, they can not achieve good optimization results for all applications. Prisacariu and others implemented the parallel version of hog algorithm under the NVIDIA CUDA architecture. For the KD tree algorithm, Zhou and others also realized the KD tree search algorithm that can meet the requirements of real-time processing by using the general graphics processor (GPGPU). Irschara, Christopher and others have accelerated optimization for the query part of VOC tree yellow method.

Based on the current research, a research on the retrieval and characteristics of multimedia testing machine is proposed. This paper mainly focuses on the analysis and research of media content of multi-media integration in the field of multimedia information management. From the above related research and the analysis of the main research problems in this field, it can be seen that although there are many studies on media content analysis, most of the multimedia fusion analysis is only in the exploration stage for specific applications, lack of a consistent conceptual framework, and lack of theoretical induction and summary of media content analysis methods of multi-media fusion. In addition, the content analysis methods of composite media, especially the content analysis of film and television programs, can be further studied from different angles and methods to obtain better analysis results[7].

Retrieval and feature analysis of multimedia testing machine integrating Bragg equation
Multimedia Fusion Analysis and retrieval method

At present, Multimedia Fusion Analysis and retrieval methods are divided into three categories: Single media cross fusion index, single media result fusion and multi-media feature fusion. The three methods have their own advantages and disadvantages for different applications. In order to achieve maximum performance in specific applications, these fusion technologies may be mixed together. Due to technical limitations, single medium cross index and single medium result fusion technology are widely used at present.

Multimedia features
Image features
Color characteristics

For objects and regions in an image, color feature is one of the main visual perception of the object or region. Typical color features include color histogram features, color moment features, color correlation vector features and so on.

Texture features

Texture features represent the basic structure of vision, especially complex and exquisite basic structure or composition, or surface appearance and surface feeling, uneven or rough surface characteristics. Texture features include roughness, directivity, contrast, periodicity, convexity and concavity and so on. Many images in nature and artificial works have obvious texture features. Typical texture features include Tamura texture features, wavelet texture features, co-occurrence autoregressive texture features and so on.

Shape features

Shape is the surface configuration, contour or perimeter of an object. The shape of an object is distinguished from its surrounding objects by its contour and shape. Shape features include image object boundary, boundary inflection point, shape center of gravity and each order moment. In addition, the outline of an object is also a shape feature.

Spatial constraints

It mainly focuses on the spatial relationship and topological relationship of objects in the image. This relationship includes the orientation relationship, coverage relationship and inclusion relationship of two or more objects.

Audio features

Audio is a non-stationary random process, and its characteristics change with time, but this change is very slow. In view of this, the audio signal can be divided into some successive short segments for processing. This is short-term processing technology. These short segments are generally 10 ~ 20ms long and are called frames. Note that the concept of frames here is different from that in video stream. We call them audio frames. Adjacent frames can be partially overlapped, and each frame can be regarded as intercepted from a continuous audio with fixed characteristics. This continuous audio is usually considered to be obtained by the periodic repetition of the short segment of audio. Therefore, processing each short segment of audio is equivalent to processing one cycle of continuous audio, or processing continuous audio with fixed characteristics[8].

Multimedia retrieval system

Traditional text retrieval systems usually include the extraction of keywords / phrases, the establishment of keyword / phrase index structure, the calculation of document similarity and the final optimization of query results. The traditional multimedia retrieval system is based on the annotation or title of relevant multimedia materials through text retrieval. If the annotation or title is inaccurate or even completely irrelevant, the accuracy of retrieval will be greatly reduced. Therefore, now the mainstream multimedia retrieval system is developed into content-based retrieval. The retrieval process mainly depends on the feature information extracted from pictures / videos, rather than related text materials.

Design and introduction of multimedia retrieval test set

Due to the large amount of computation and data, the processing speed of multimedia retrieval algorithm is greatly limited, so it can not meet the real-time needs of users. The existing architecture can not have high efficiency for this kind of data intensive and computing intensive applications. Therefore, on the one hand, we need to reduce the amount of calculation of image retrieval algorithm from the algorithm level and improve the processing speed. On the other hand, we need to design a new architecture to make this kind of algorithm reach the real-time processing requirements with low energy consumption and high efficiency, so that the multimedia retrieval algorithm can be effectively used in practice.

Whether we optimize and accelerate the algorithm or design a new architecture, we need to have a comprehensive and in-depth understanding of the image retrieval algorithm. To fully understand the characteristics of image retrieval algorithm, there must be a representative test set for research. Therefore, this paper designs a benchmark program test set for image retrieval applications, including the common algorithms in each main stage of image detection algorithm.

Methodology of Multimedia Fusion Analysis

The semantic understanding of complex media itself is very complex. Generally, there are many kinds of media semantics to be processed in media content analysis. Even if we can completely obtain various media features that can accurately describe media semantics, it is still difficult to correctly understand media semantics. For example, the dialogue scene in film and television may be the dialogue between two people, the dialogue between three people, or even the dialogue between more people; Even the dialogue between two people is very different in film and television performance. Therefore, media content analysis cannot be an understanding of the semantics of the fixed form of media, nor can it be a simple matching of a certain feature. Second, the difference between different media semantics lies not only in their corresponding feature values, but also in the characteristics used to define them; The methods of using these features in media content analysis may also be different. This makes it difficult to classify various media semantics accurately and reliably[9].

There are two basic ideas to solve this problem. One is to find - an “all inclusive” media semantic representation method and a matching scheme that can understand the semantics of each class. Such methods may be effective on some specific problems, but at present, no method has shown the ability to understand the semantics of multiple media. The other is a knowledge-based method, which uses knowledge to combine all processing methods for specific media semantics: the model of each type of media semantics is defined by various features and their variation range for the members of this class, as long as these features have corresponding processing methods to detect or measure. There is relevant “control knowledge” for the media semantics of each category, select which features to use in processing and the order in which these features are used, and combine the results of each part matching processing. As shown in Figure 1.

Figure 1

Analysis steps of Multimedia Fusion

Application of Multimedia Fusion retrieval technology

In most multimedia applications, structured data, text, audio, image and other media appear in the form of a combination of integrated information units, rather than alone, which is called unit multimedia object (M Mo). According to the difficulty of processing, the media can be classified into structured data, free text, audio, image and video. Some relatively easy to handle media types are usually added to help index and retrieve difficult media. For example, free text is usually added to images and videos in the form of annotations to better use the information contained in these media. In order to index and retrieve MMO effectively, each contained media and their temporal, spatial and semantic relationships can be utilized. Different media composition can have different MMO types. The following are three common MMO types with member data:

Type 1: Structured attribute; Free text; Audio.

Type 2: Structured attribute; Free text; Audio; Image.

Type 3: Structured attribute; Free text; Audio; Video.

Significance of Bragg equation

The physical meaning of Bragg equation is not very clear. It is only a mathematical treatment of Laue equation, which shows that the lattice structure reflected by Laue diffraction just corresponds to the reflection of its reciprocal lattice, which is not necessarily the real lattice. Because in the Bragg equation, the distance between the crystal planes a is a harsh limiting condition. It only represents the distance between the two crystal planes, but it does not mean that the two ends of a will fall on the lattice points of the two crystal planes[10].

Bragg equation is an important equation in crystal chemistry. It connects crystal diffraction with crystal lattice. Because crystal diffraction is a scattering phenomenon of X-rays by electrons in crystals. It is not a direct projection of atoms or molecules, so there is no obvious and direct geometric image correspondence between diffraction and crystal lattice, and lattice transformation is needed. From the Laue equation to the Bragg equation, the geometric relationship between this diffraction and the lattice will be clear with the help of the inverse lattice.

The vector form of Laue equation is shown in formula (1) (2) (3): a=(SSo)=hλSSoλ=h1a=h1a=hax a = \left( {S - So} \right) = h\lambda {{S - So} \over \lambda } = h \bullet {1 \over a} = h{1 \over a} = h \bullet {a^x} b(SSo)=kλSSoλ=k1b=kbx b\left( {S - So} \right) = k\lambda {{S - So} \over \lambda } = k \bullet {1 \over b} = k{b^x} c=(SSo)=1λSSoλ=11c=1cx c = \left( {S - So} \right) = 1\lambda {{S - So} \over \lambda } = 1 \bullet {1 \over c} = 1{c^x}

They are inverse vectors of vectors a, b and c, or reciprocal vectors of each other.

Experiment and analysis

The design and implementation of the system of this subject is mainly divided into three layers: the interaction layer, the business logic layer and the database server layer. The interaction layer mainly realizes the interaction between the system and the user, obtains the retrieval requirements and feeds back the retrieval results; the business logic layer mainly analyzes the input information, processes the data, and realizes the interaction with the database; the database server layer mainly realizes the search and retrieval of data. This section presents the implementation of the entire retrieval process.

(1) Static property retrieval implementation

The static attributes include image category, image format, image width, etc. These attributes are designed in the form of text boxes or selectable boxes, and support direct input of attributes or selection of interesting retrieval attributes.

(2) Dynamic attribute retrieval implementation

Image dynamic attribute retrieval refers to setting the retrieval parameters according to the sample image, and retrieving the image data that meets the similarity requirements with the sample image. This includes two processes: uploading the sample image and setting the parameters.

(3) Joint retrieval of dynamic attributes and static attributes

The joint retrieval strategy of dynamic attributes and static attributes is to realize the retrieval of heterogeneous features. The interface mainly includes three parts: dynamic attribute selection area, static attribute value input area, and parameter setting area. When searching, you can flexibly choose search attributes and search strategies. Among them, dynamic attributes can be retrieved by selecting color features, shape features, and texture features as content alone, or a retrieval strategy based on comprehensive features can be used for comprehensive retrieval of the three features. Static property settings are also diverse and can be retrieved according to different content[11].

(4) Manually assisted selection area to realize retrieval

Considering that the system may only be interested in a certain area or an object in the image during retrieval, and hope to retrieve an image similar to the object or area, the retrieval area needs to be manually selected with human assistance. The white dotted border on the left side of the figure is the selection box. Select the area by dragging the border, and change the size of the border to determine the size of the area. The right part is a preview of the selected area, and a text box below the image shows the size of the selected area.

Architecture design of Multimedia Fusion Analysis System

The design process of multimedia fusion analysis system architecture is to transform the functional model of multimedia fusion analysis system introduced in the previous section into the actual system structure. The process follows the usual information system design sequence: requirement design → functional design → physical design + system evaluation. This design process is a multiple mapping process. As shown in Figure 2, mapping transforms the system requirements into the functional set to realize these requirements, and mapping II transforms the functional set into various hardware or software sets to complete these functions, which form the architecture of the multimedia fusion analysis system.

Figure 2

Process of architecture design of Multimedia Fusion Analysis System

(1) Static property retrieval implementation

Video static attributes refer to general attributes that people are interested in and can be used for retrieval. The static attributes of videos include video category, format, author, shooting location, name, subject, descriptive content, etc. Select the video content attribute of interest and enter the attribute value to retrieve videos that meet the requirements of the attribute value[12].

(2) Dynamic attribute retrieval implementation

The video dynamic attribute retrieval implementation is based on key frame retrieval. Taking a key frame as a retrieval example, the process of finding videos in the video library that meet the similarity requirement with the key frame is similar to the image retrieval interface, including two processes: uploading key frames and setting parameters.

Comprehensive image indexing and retrieval

The case of integrated image indexing and retrieval is similar to that of audio indexing and retrieval. Images are indexed and retrieved based on low-level features. Structured attributes and free text annotations can be used for image indexing and retrieval. The main factor of the characteristics of the field distribution function is that the farther away from the heat source, the greater the influence on the lag time and time constant of the temperature field, while the position of the heat source has little influence on the gain coefficient of the three-dimensional temperature field of the cylinder. As shown in Figure 3.

Figure 3

Variation diagram of temperature field

As can be seen from Figure 4, there are certain differences in the retrieval performance of different image categories. Since the retrieval method designed in this paper requires manual and dynamic parameter setting, different categories and different users have different retrieval priorities. Therefore, Min The value setting and the setting of the retrieval parameters of each feature component have an impact on the retrieval system of the system. At the same time, the evaluation method of this paper relies on manual statistics, so there will be certain errors in the performance evaluation[13].

Figure 4

Retrieval Performance Comparison - Precision Comparison

As a medium for carrying information, multimedia data occupies an increasing proportion in the Internet, and the demand for query and retrieval of multimedia information is becoming more and more urgent. The content-based retrieval technology of multimedia data is a hot issue in current research[14]. Different from traditional character and numerical data, multimedia data is unstructured data, and the management capability of database to multimedia data needs to be expanded. There are many types of multimedia data, and the content characteristics of various multimedia data are also different.

Conclusion

This paper discusses the media content analysis of Multimedia Fusion in theory, and discusses some basic problems involved in multimedia fusion analysis. Aiming at the basic problems faced by media content analysis, this paper formally describes the media content analysis method of multimedia fusion, and points out that the knowledge-based model driven method is an effective method to solve the problem of multimedia fusion analysis. It also discusses the architecture of multimedia fusion analysis system, and summarizes the basic components of Multimedia Fusion Analysis System: media processing subsystem, auxiliary information processing system, multimedia fusion processing subsystem, man-machine interface, storage management subsystem and data network. Based on the establishment of the conceptual model of media content analysis, two kinds of media feature fusion analysis models, conditional constraint model and evidence combination model, are established. The characteristics of the two models are analyzed, and it is pointed out that the evidence combination model has more advantages than the conditional constraint model. Finally, some related problems in multimedia fusion analysis are discussed. As the basis of the design experiment, this paper further studies the architectural features of the chosen algorithms, in order to obtain some information about the architectural design and implementation. for better efficiency associated with the use of multiple data retrieval. Based on the features listed above, this document also releases some of the ultimate hardware design and optimization of the end, including single-node processor design, built in one round equations, and simulation and evaluation using this experiment. 's suggestion.

Figure 1

Analysis steps of Multimedia Fusion
Analysis steps of Multimedia Fusion

Figure 2

Process of architecture design of Multimedia Fusion Analysis System
Process of architecture design of Multimedia Fusion Analysis System

Figure 3

Variation diagram of temperature field
Variation diagram of temperature field

Figure 4

Retrieval Performance Comparison - Precision Comparison
Retrieval Performance Comparison - Precision Comparison

Yal A M, Ekici M. Stationary optical solitons with complex Ginzburg–Landau equation having nonlinear chromatic dispersion[J]. Optical and Quantum Electronics, 2022, 54(3). YalA M EkiciM Stationary optical solitons with complex Ginzburg–Landau equation having nonlinear chromatic dispersion [J] Optical and Quantum Electronics 2022 54 3 10.1007/s11082-022-03557-3885328935194297 Search in Google Scholar

M Jiménez, Aguilar J, Monsalve-Pulido J, et al. An automatic approach of audio feature engineering for the extraction, analysis and selection of descriptors[J]. International Journal of Multimedia Information Retrieval, 2021, 10(1):33–42. JiménezM AguilarJ Monsalve-PulidoJ An automatic approach of audio feature engineering for the extraction, analysis and selection of descriptors [J] International Journal of Multimedia Information Retrieval 2021 10 1 33 42 10.1007/s13735-020-00202-1 Search in Google Scholar

Cheng Q, Gu X. Bridging multimedia heterogeneity gap via Graph Representation Learning for cross-modal retrieval[J]. Neural Networks, 2021, 134(3):143–162. ChengQ GuX Bridging multimedia heterogeneity gap via Graph Representation Learning for cross-modal retrieval [J] Neural Networks 2021 134 3 143 162 10.1016/j.neunet.2020.11.01133310483 Search in Google Scholar

Suresha M, Kuppa S, Raghukumar D S. A study on deep learning spatiotemporal models and feature extraction techniques for video understanding[J]. International Journal of Multimedia Information Retrieval, 2020, 9(2):81–101. SureshaM KuppaS RaghukumarD S A study on deep learning spatiotemporal models and feature extraction techniques for video understanding [J] International Journal of Multimedia Information Retrieval 2020 9 2 81 101 10.1007/s13735-019-00190-x Search in Google Scholar

Bekhet S, Ahmed A. Evaluation of similarity measures for video retrieval[J]. Multimedia Tools and Applications, 2020, 79(9–10):6265–6278. BekhetS AhmedA Evaluation of similarity measures for video retrieval [J] Multimedia Tools and Applications 2020 79 9–10 6265 6278 10.1007/s11042-019-08539-4 Search in Google Scholar

Tripathy S K, R Srivastava. AMS-CNN: Attentive multi-stream CNN for video-based crowd counting[J]. International Journal of Multimedia Information Retrieval, 2021, 10(4):239–254. TripathyS K SrivastavaR AMS-CNN: Attentive multi-stream CNN for video-based crowd counting [J] International Journal of Multimedia Information Retrieval 2021 10 4 239 254 10.1007/s13735-021-00220-7 Search in Google Scholar

Fadel S G, Torres R. Neural relational inference for disaster multimedia retrieval[J]. Multimedia Tools and Applications, 2020, 79(35):26735–26746. FadelS G TorresR Neural relational inference for disaster multimedia retrieval [J] Multimedia Tools and Applications 2020 79 35 26735 26746 10.1007/s11042-020-09272-z Search in Google Scholar

Cevik N, Cevik T. A novel high-performance holistic descriptor for face retrieval[J]. Pattern Analysis and Applications, 2020, 23(1):371–383. CevikN CevikT A novel high-performance holistic descriptor for face retrieval [J] Pattern Analysis and Applications 2020 23 1 371 383 10.1007/s10044-019-00803-5 Search in Google Scholar

Zhang L, He Z, Yang Y, et al. Tasks Integrated Networks: Joint Detection and Retrieval for Image Search[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, PP(99):1–1. ZhangL HeZ YangY Tasks Integrated Networks: Joint Detection and Retrieval for Image Search [J] IEEE Transactions on Pattern Analysis and Machine Intelligence 2020 0 PP 1 1 10.1109/TPAMI.2020.300975832750818 Search in Google Scholar

Xu C, Leng B, Chen B, et al. Learning Discriminative and Generative Shape Embeddings for Three-Dimensional Shape Retrieval[J]. IEEE Transactions on Multimedia, 2020, 22(9):2234–2245. XuC LengB ChenB Learning Discriminative and Generative Shape Embeddings for Three-Dimensional Shape Retrieval [J] IEEE Transactions on Multimedia 2020 22 9 2234 2245 10.1109/TMM.2019.2957933 Search in Google Scholar

M Jiménez, Aguilar J, Monsalve-Pulido J, et al. An automatic approach of audio feature engineering for the extraction, analysis and selection of descriptors[J]. International Journal of Multimedia Information Retrieval, 2021, 10(1):33–42. JiménezM AguilarJ Monsalve-PulidoJ An automatic approach of audio feature engineering for the extraction, analysis and selection of descriptors [J] International Journal of Multimedia Information Retrieval 2021 10 1 33 42 10.1007/s13735-020-00202-1 Search in Google Scholar

Ahmed K T, Aslam S, Afzal H, et al. Symmetric Image Contents Analysis and Retrieval using Decimation, Pattern Analysis, Orientation, and Features Fusion[J]. IEEE Access, 2021, PP(99):1–1. AhmedK T AslamS AfzalH Symmetric Image Contents Analysis and Retrieval using Decimation, Pattern Analysis, Orientation, and Features Fusion [J] IEEE Access 2021 PP 99 1 1 10.1109/ACCESS.2021.3071581 Search in Google Scholar

Ilknur Koca and Pelin Yaprakdal. A new approach for nuclear family model with fractional order Caputo derivative[J]. Applied Mathematics and Nonlinear Sciences, 2020, 5(1) : 393–404. KocaIlknur YaprakdalPelin A new approach for nuclear family model with fractional order Caputo derivative [J] Applied Mathematics and Nonlinear Sciences 2020 5 1 393 404 10.2478/amns.2020.1.00037 Search in Google Scholar

Saouli M A. Existence of solution for Mean-field Reflected Discontinuous Backward Doubly Stochastic Differential Equation[J]. Applied Mathematics and Nonlinear Sciences, 2020, 5(2):205–216. SaouliM A Existence of solution for Mean-field Reflected Discontinuous Backward Doubly Stochastic Differential Equation [J] Applied Mathematics and Nonlinear Sciences 2020 5 2 205 216 10.2478/amns.2020.2.00038 Search in Google Scholar

Articoli consigliati da Trend MD

Pianifica la tua conferenza remota con Sciendo