The underwater digital images generally suffer from blur, low contrast, non uniform lighting and diminished color.This preprocessing technique based on image to improve the quality of underwater digital images.The mixed Contrast Limited Adaptive Histogram Equalization (CLAHE) has neglected the utilization of L*A*B* color image space to improve the image in an effective way.Also the uneven illumination problem is ignored by many researchers.To conquer the problems of uneven illumination in the resultant image of the CLAHE image output has been further removed by utilizing the smoothing process of image gradient.The main objective is to enhance the accuracy of the underwater digital image enhancement techniques.Various types of digital images will be considered for experimental point of view to estimate the efficiency of the image enhancement methods or techniques.Also, various types of image top-quality metrics have been utilized in order to check the significant improvement of the recommended technique over the offered techniques.The significant improvements have shown in the comparative analysis of the proposed algorithm over the available mixed CLAHE
Inscriptions are the part of history all over the world.The information on the inscriptions are very important to the mankind, but understanding and transforming it is need of today as epigraphists are in extinct condition.Many researches are voted for the restoration, segmentation, and classification of such inscriptions and it is still in progress.In making an attempt, a transformation system is proposed which Normalizes, Segments and classifies the characters on the inscriptions to modern readable characters.
Electrocardiogram (ECG) is a non-invasive technique which is used as a main diagnostic device for cardiovascular diseases.A cleaned ECG signal offers essential information for electrophysiology for the heart diseases and ischemic changes that might happen.It gives significant information for the functional aspects of the heart and cardiovascular system.In this research, the detection of cardiac arrhythmias in the ECG signal consists of: pre-processing using DWT, detection of QRS complex in ECG signal; feature extraction from detected QRS complexes; classification of beats using extracted feature set from QRS complexes.To see the presence of cardiovascular disease in the Rpeaks, GA (Genetic algorithm) and NN (Neural network) are considered.Initially, the extraction of the R peaks are done precisely through feature extraction and then they are optimized to have the small value of actual R peaks through which the user can view the presence of disease with the variation in the beat.To check the performance of the proposed work, varied parameters, viz.Precision, recall, F-measure, Accuracy, Error and time are analyzed and calculated.
In this paper the effects of two identical notches in the same non radiating edge and opposite non radiating edges have been studied extensively.The effect on resonant frequency and gain with the variation of the notch length and notch width are studied.It has been observed that the notches in the opposite non radiating edges affect more on resonant frequency than the notches in the same non radiating edge of a microstrip patch antenna.About 80.2% size reduction has been achieved by the dual notches in the same non radiating edge and about 87.5% size reduction is achieved for the dual notches in the opposite non radiating edges.
To determine the location of the acoustic emission source is one of the main purpose of the acoustic emission test.The accuracy of positioning reflects the degree of coincidence between the position of the sound source and the source location of the actual active defects.Time difference location method and zone location method are the main methods of acoustic emission source location.However, the time difference positioning method has high reliability, and therefore, most of the tests and acoustic emission instruments use time difference method to locate the source of acoustic emission signal.Three point time difference position method and the non-iterative multipoint position method are employed to locate experiment to different types of acoustic emission signals collected.Based on the positioning results, three point time difference positioning method and the non-iteration multipoint positioning method can better position acoustic emission source.However, when the sensor receives the small signal time difference, three point time difference positioning method has large error.The non-iteration multipoint positioning method has positioning advantage when more than four sensors can receive an event signal.
With the advancement of modern technology, imaging has become a significant inventory with a great range of applications throughout medical science, biological research and industrial safety; if analyzed properly.However, considering the flexibility and appropriateness for varying applications most of the image processing methods do not fit in as suitable.Considering simplicity, widespread applicability and customization options image analysis using MATLAB-based custom codes has already gained popularity in many perspectives.One of the most promising applications where image analysis can cater the requirements of the user faster and with a lesser investment would be Geographical surveying.This work is based on statistical analysis of satellite map using MATLAB to exhibit how a customized application can perform separate clear mapping of various objects and areas, specifications of different terrain properties and statistical surveying of the region of interest.This program takes image from satellite map as input, segments the image in various maps like road map, field map etc. and provides necessary statistical data for each segments.It can be developed to address any further need as well as can be applied in different color-based image analysis applications if required.However, coding knowledge and techniques being unfamiliar to many people, might lessen its user-friendliness too some extent.To overcome this limitation a graphical interface of the program has been developed for the users possessing no MATLAB or computational language experience.The program is also customized to work on satellite map snapshots in different zooms.Wide range of applications of the program include geographical survey, surveillance and tracking, urban planning and design etc.
With the development of hyperspectral remote sensing classification technology, how to obtain higher classification accuracy under the condition of small samples has become a difficult and hot spot in the field of hyperspectral remote sensing technology.In recent years, Cloud model has been used in hyperspectral data analysis.In this paper, a method of hyperspectral remote sensing image classification based on EMAP (Extended Multi-Attribute Profile) and cloud model is proposed.Firstly, EMAP is a kind of texture feature which is obtained by a series of attribute filters filtering hyperspectral image, and is fused with spectral feature.The fused feature is processed by LDA to reduce the dimension, and the data of dimension reduction is taken as the final classification data.Finally, according to the training sample set, the inverse cloud generator is used to generate the multi dimension cloud model for each object, and different scales are added into different cloud models to form multi-scale cloud model.Then, the membership degree of each pixel is calculated by multi-scale cloud model.Finally, we use the maximum decision method to classify each test sample.The experimental results show that the method is simple and efficient, and the classification accuracy is improved effectively.
Monitoring of respiration is crucial for determining a patient´s health status, specially previously and after an operation.However, many conventional methods are difficult to use in a spontaneously ventilating patient.This paper presents a method for estimating respiratory rate from the signal of a photoplethysmograph.This is a non-invasive sensor that can be used to obtain an estimation of beats per minute of a given patient by measuring light reflection on the patient's blood vessel and counting changes in blood flow.The PPG signal also offers information about respiration, so respiratory rate can be obtained through signal processing.The proposed method based on digital filtering was implemented in a wearable device and tested on 30 volunteers, and the results were compared with the ones measured by traditional ways.The results show that there is no statistically significant difference between the data measured by the device and the traditional method.
Internet is very feasible and an excellent distribution system. One way to protect multimedia data against risk of illegal recording and malicious retransmission is to embed a signal, called watermark that authenticates the owner of the data. Watermark is an image or text to be embedded into the documents that needs protection. A good watermarking scheme is the one which can offer resistance against attacks and is imperceptible to hackers. Robustness measures the withstanding capability of watermark against various attacks. This paper is an effort to study different watermarking techniques on the basis of given parameters. For the purpose of implementation, MATLAB is used and comparison is done on the basis of various performance metrics such as PSNR, MSE, Time Complexity, Correlation Co-efficient and robustness against various attacks like cropping, Gaussian noise, salt and pepper noise etc. The result shows that frequency domain techniques are comparatively more robust with DWT being the most robust one.
An effective compression algorithm of hyperspectral image based on discrete wavelet transform (DWT) and improved Karhunen-Loeve transform(KLT) is proposed in this paper. It could achieve better performance with effective combination of the two methods which convert the energy of images to a small number of coefficients. In order to reduce the most redundancy of space, it deals with each spectrum with fast 5/3 2-D DWT firstly. Then the spectral 1-D KLT is applied on the 2-D DWT coefficients to reduce spectral redundancy and the other spatial redundancy. Finally, entropy coding is executed to obtain the compressed code stream. The experimental results show that the average peak signal to noise ratio (PSNR) of the proposed compression algorithm is 3.01 bit per pixel (bpp), which improves greatly through comparing with the other approaches, and reduces the operating time and improves the performance of hyperspectral image compression algorithm. It proposes the hardware implementation strategy at the same time, verifies the correctness and feasibility of our method. The work in this paper has a good reference value and significance for the similar case.
We introduce in the paper a method for completing Magnetic Resonance Image(MRI) feature processing by multiresolution representations.The image is decomposed using dyadic wavelet transformation.A new threshold is proposed by way of different thresholds according to the different scales of wavelet coefficients to deal with the wavelet coefficients.The nonlinear enhancing operator is applied for wavelet coefficients with the enhancement method in the corresponding scales.Our results show that the proposed method outperforms conventional enhancement method, the information entropy and contrast improvement index of the images is improved, enhances edge texture of images.
As the wide use of digital devices and Internet-based applications, the amount of data on the whole world is growing significantly.For facing the massive of data sets like texts, videos, images, GPS, even medical information, data mining would be a suitable approach to make full use of the data for supporting advanced decision-makings.This paper introduces a rough set theory-based data mining approach for massive data sets.This approach uses an innovative discretization method to make the decision table to be compatible.Three steps with suitable models or algorithms are equipped to the clustering method.Based on the clustered data, the rough set-enabled data mining approach is introduced.Experiments are carried out through three levels to test the feasibility of the proposed approach by comparing the Greedy algorithm, information entropy-based algorithm, and importance-based algorithm.Several contributions are significant from this paper.Firstly, it could be observed that after using the proposed approach, the amount of breakpoints candidates has been greatly reduced.Some experiments achieve about 85% reduction.The reason is attributed to the Step 3 which makes the time complexity achieve. Secondly, it could be observed that the proposed approach outperforms the other three methods in the computational time.As the increasing of breakpoints candidates, the advantages of the proposed algorithm are more obvious.Finally, the information entropy-based approach (III) is prone to memory overflow.That means this approach takes much more memory to carry out the calculation.
This paper describes development and usability evaluation of Korea's Hangeul font registration system.As the use of fonts increases, various fonts are produced.However, finding the fonts you want is not an easy task.Therefore, we have developed a system that can integrate existing fonts and search for desired fonts through the Korean font registration system.In addition, we evaluated the possibility and improvement plan of the system by evaluating whether the system can be actually useful to users.
Many kinds of transportation system have been developed to help the people to move things or people. This paper presents an enhanced algorithm for the detection rate to classify the two-wheelers with riding people using new innovative global and local feature extraction and comparison method; the difference of Gaussian (DoG), the normalized cross correlation (NCC) and its application of histogram of oriented gradients (HOG). Applied new features are calculated by DoG which present edge components and NCC algorithm which is used to make a match from template images, local weighting values are calculated from template cell features. And the combined cell features are classified by using a boosting algorithm (Adaboost). The improvement detection rates are confirmed through experiments using bicycles and motorcycles data set for 60 and 90 degrees.
The x-ray image processing is an effective technique to distinguish between the major types of arthritis: Osteoarthritis Arthritis (OA) and rheumatoid Arthritis (RA). The degenerative bone disorders are diagnosed using X-ray detection. X-ray scans alone are insufficient to detect the type of arthritis. Image processing can aid in improving the diagnosis. The classification is done on the basis of differentiation in the region properties and boundary areas that can be identified using MATLAB. These properties were used to better understanding the variation in the knee, hands and neck region. The study holds potential as a diagnostic tool for arthritis identification through x-rays. It could pave way to differential diagnosis in future.
At fully mechanized caving mining face, the degree of coal falling is mainly determined by worker's experience, which can easily lead to the problem of "over caving" or "under caving".Aiming at the problem, this paper put forward a method of coal and rock recognition based on variational mode decomposition and PSO-SVM.This paper carried out the top coal and rock caving experiment and obtained the vibration signal of the hydraulic support tail beam under the different working conditions of coal falling and rock falling.We analyzed the signal with the method of variational mode decomposition after the number of Intrinsic Mode functions confirmed.Through comparative analysis, we found that there are many differences in modal function energy percentage among different work conditions.Mode energies of coal signal are mainly distributed in 0 ~ 1500Hz, which accounts for about 50% of the total energy.Mode energies are mainly distributed in 3000 ~ 5000 Hz, which accounts for more than 50% of the total energy.The results indicate that the energy percentage of modal functions can be used as the characteristic index for the coal and rock character recognition at fully mechanized caving mining face.Then we used the PSO-SVM algorithm to do the classification, and the classification accuracy of this method is 90%.The conclusion indicates that the method put forward can be well applied in realizing coal and rock identification.
Target location is an effective approach to mass information filtering for remote sensing images.In this paper, a simple and fast target location method is proposed based on line features for pre-registrated remote sensing images.Firstly, a line model of the specific target to be located is generated by manually labeling line features in the historical images containing the target.Secondly, straight lines are extracted from the remote sensing images to be searched.Thirdly, the target is located by matching line features in the line model and extracted lines in the remote sensing image.The optimal matching position is obtained by offset-accumulations similar with a voting mechanism.Numerous experimental results reveal that our proposal is effective and efficient for target locating with pixel-level accuracy in remote sensing images.Our proposal is also robust for partially occlusion in remote sensing images with small registration error.
Lung cancer is one of the most prevalent causes of death among cancer patients. Later thee diagnosis is, lesser are the chances of recovery. However, patients have a higher chance of survival if it can be diagnosed in the primary stages. No one can reject the importance of CT, CT-PET & PET modalities in the diagnosis of pulmonary problems. Modalities mentioned above have their own limitations. Digital Image Processing Techniques can be helpful in detection of lung nodule. Utilization of various imaging techniques is helpful in ruling out the area and type of anomaly. Contrast Adjustment, Segmentation, Thresholding of image are such techniques which enhance the acuity of the visibility of the defected region, however the feature extraction methodologies are helpful in ruling out the specificity of the defected region determining the extent of damage to the soft tissue structures. Utilization of such methods in routine clinical practices can serve as a helping hand to clinicians, as they will be able to take decision of treatment earlier with early detection of anomalies s. In this paper we developed such methodology which can be helpful in early detection and diagnosis of the presence on lung anomaly.
As a kind of important network resources, Flash animation has been widely studied, but its network retrieval has a lot of problems.In order to improve the retrieval accuracy and fast preview the search results, we mainly study the process of extraction of Flash animation visual scene representative frame.We proposed an adaptive threshold segmentation algorithm of the Flash animation visual scene, and extraction algorithm of representative frame based on the average main color distance of the key frame.We download 82 Flash animation samples for visual scene segmentation from the Internet, and then we use the correct visual scenes to extract their representative frame.Finally, we follow the classification of MTV, games, cartoons, and advertising to analyze the segmentation results of each type of Flash animation visual scenes and the representative frame extraction effect.The visual scene representation frame obtained in this study can be used to make the dynamic summary of Flash animation in order to preview the search results of network Flash resource search, and our research results have important implications for the Flash resource retrieval based on content.
