Semantic Segmentation by Using Down-Sampling and Subpixel Convolution: DSSC-UNet

Recently, semantic segmentation has been widely applied toimage processing, scene understanding, and many others. Especially, indeep learning-based semantic segmentation, the U-Net with convolutionalencoder-decoder architecture is a representative model which is proposed forimage segmentation in the biomedical field. It used max pooling operationfor reducing the size of image and making noise robust. However, instead ofreducing the complexity of the model, max pooling has the disadvantageof omitting some information about the image in reducing it. So, thispaper used two diagonal elements of down-sampling operation instead ofit. We think that the down-sampling feature maps have more informationintrinsically than max pooling feature maps because of keeping the Nyquisttheorem and extracting the latent information from them. In addition,this paper used two other diagonal elements for the skip connection. Indecoding, we used Subpixel Convolution rather than transposed convolutionto efficiently decode the encoded feature maps. Including all the ideas, thispaper proposed the new encoder-decoder model called Down-Sampling andSubpixel Convolution U-Net (DSSC-UNet). To prove the better performanceof the proposed model, this paper measured the performance of the UNetand DSSC-UNet on the Cityscapes. As a result, DSSC-UNet achieved89.6% Mean Intersection OverUnion (Mean-IoU) andU-Net achieved 85.6%Mean-IoU, confirming that DSSC-UNet achieved better performance.

铁汉柔情——音乐传真A3.2 24bit Up-sampling CD播放机

面前 这台A3.2合并式 CD播放机,是英国Musi cal Fidelity(音乐传真)公司 新近投放市场的中高级机种。在过去 的三年里,其前身A3 CD机凭借着优异的综合 实力已成为同级产品中的佼佼者。那么A3.2能否取 得同样的成就呢? 如今,音乐传真的产品不论是CD机还是放大器,

Up-Sampled Cross-Correlation Based Object Tracking & Vibration Measurement in Agriculture Tractor System

This research introduces a challenge in integrating and cleaning the data,which is a crucial task in object matching.While the object is detected and then measured,the vibration at different light intensities may influence the durability and reliability of mechanical systems or structures and cause problems such as damage,abnormal stopping,and disaster.Recent research failed to improve the accuracy rate and the computation time in tracking an object and in the vibration measurement.To solve all these problems,this proposed research simplifies the scaling factor determination by assigning a known real-world dimension to a predetermined portion of the image.A novel white color sticker of the known dimensions marked with a color dot is pasted on the surface of an object for the best result in the template matching using the Improved Up-Sampled Cross-Correlation(UCC)algorithm.The vibration measurement is calculated using the Finite-Difference Algorithm(FDA),a machine vision systemfitted with a macro lens sensor that is capable of capturing the image at a closer range,which does not affect the quality of displacement measurement from the video frames.Thefield test was conducted on the TAFE(Tractors and Farm Equipment Limited)tractor parts,and the percentage of error was recorded between 30%and 50%at very low vibration values close to zero,whereas it was recorded between 5%and 10%error in most high-accelerations,the essential range for vibration analysis.Finally,the suggested system is more suitable for measuring the vibration of stationary machinery having low frequency ranges.The use of a macro lens enables to capture of image frames at very close-ups.A 30%to 50%error percentage has been reported when the vibration amplitude is very small.Therefore,this study is not suitable for Nano vibration analysis.

Scalable transcoding for video transmission over space-time OFDM systems

A new scheme combining a scalable transcoder with space time block codes （STBC） for an orthogonal frequency division multiplexing （OFDM） system is proposed for robust video transmission in dispersive fading channels. The target application for such a scalable transcoder is to provide successful access to the pre-encoded high quality video MPEG-2 from mobile wireless terminals. In the scalable transcoder, besides outputting the MPEG-4 fine granular scalability （FGS） bitstream, both the size of video frames and the bit rate are reduced. And an array processing algorithm of layer interference suppression is used at the receiver which makes the system structure provide different levels of protection to different layers. Furthermore, by considering the important level of scalable bitstream, the different bitstreams can be given different level protection by the system structure and channel coding. With the proposed system, the concurrent large diversity gain characteristic of STBC and alleviation of the frequency-selective fading effect of OFDM can be achieved. The simulation results show that the proposed schemes integrating scalable transcoding can provide a basic quality of video transmission and outperform the conventional single layer transcoding transmitted under the random and bursty error channel conditions.

Low Bit Rate Underwater Video Image Compression and Coding Method Based on Wavelet Decomposition

In view of the limited bandwidth of underwater video image transmission,a low bit rate underwater video compression coding method is proposed.Based on the preprocessing process of wavelet transform and coefficient down-sampling,the visual redundancy of underwater image is removed and the computational coefficients and coding bits are reduced.At the same time,combined with multi-level wavelet decomposition,inter frame motion compensation,entropy coding and other methods,according to the characteristics of different types of frame image data,reduce the number of calculations and improve the coding efficiency.The experimental results show that the reconstructed image quality can meet the visual requirements,and the average compression ratio of underwater video can meet the requirements of underwater acoustic channel transmission rate.

A Universal Approach to Designing an Image Interpolator with an Image Smoothing Filter

A number of conventional interpolation techniques have been proposed. However, it seems that there do not exist good criteria for the design of optimal linear interpolators. Also, such an interpolator can hardly provide a satisfactory solution for interpolating noisy images. In this paper, the novelty of this research is that a universal approach is proposed to design an image interpolator with any one image smoothing filter, thereby not only interpolating a down-sampled image but also preserving the characteristics of the performing filtering.

Component-adaptive up-sampling for inter layer interpolation in scalable video coding

Scalable video coding （SVC） is a newly emerging standard to be finalized as an extension of H.264/AVC. The most attractive characters in SVC are the inter layer prediction techniques, such as Intra_BL mode. But in current SVC scheme, a uniform up-sampling filter （UUSF） is employed to magnify all components of an image, which will be very inefficient and result in a lot of redundant computational complexity. To overcome this, we propose an efficient component-adaptive up-sampling filter （CAUSF） for inter layer interpolation. In CAUSF, one character of human vision system is considered, and different up- sampling filters are assigned to different components. In particular, the six-tap FIR filter used in UUSF is kept and assigned for luminance component. But for chrominance components, a new four-tap FIR filter is used. Experimental results show that CAUSF maintains the performances of coded bit-rate and PSNR-Y without any noticeable loss, and provides significant reduction in computational complexity.

Dehazing for Image and Video Using Guided Filter

Poor visibility in bad weather, such as haze and fog, is a major problem for many applications of computer vision. Thus, haze removal is highly required for receiving high performance of the vision algorithm. In this paper, we propose a new fast dehazing method for real-time image and video processing. The transmission map estimated by an improved guided filtering scheme is smooth and respect with depth information of the underlying image. Results demonstrate that the proposed method achieves good dehazeing effect as well as real-time performance. The proposed algorithm, due to its speed and ability to improve visibility, may be used with advantages as pre-processing in many systems ranging from surveillance, intelligent vehicles, to remote sensing.

The Improvement of ECG Parameter Estimation Obtained by Up-sampling 500 Hz Data

Nowadays, numerous commercial electrocardiographs(ECG) have500 Hz output sampling rate, i.e., 2 ms corresponding sampling interval. Medical workers who have no background of signal processing may worry about whether such low 500 Hz output sampling rate data could support high-precision estimation of electrocardiogram parameters by being up-sampled. With intention to clarify the problem to the medical workers, this paper takes some simulating experiment to illustrate it.The experiment chose the RR interval and R peak amplitude as representative parameters to be estimated. The obtained results were as follows: for RR interval parameter,its estimation error can be lowered to below 0.3 ms by up-sampling; for R peak amplitude, its estimation error can be reduced to below 2.0 μV by up-sampling. It is believed that other ECG parameter estimation can also be improved similarly by up-sampling 500 Hz data.

Deep Convolutional Network Based Machine Intelligence Model for Satellite Cloud Image Classification

As a huge number of satellites revolve around the earth,a great probability exists to observe and determine the change phenomena on the earth through the analysis of satellite images on a real-time basis.Therefore,classifying satellite images plays strong assistance in remote sensing communities for predicting tropical cyclones.In this article,a classification approach is proposed using Deep Convolutional Neural Network(DCNN),comprising numerous layers,which extract the features through a downsampling process for classifying satellite cloud images.DCNN is trained marvelously on cloud images with an impressive amount of prediction accuracy.Delivery time decreases for testing images,whereas prediction accuracy increases using an appropriate deep convolutional network with a huge number of training dataset instances.The satellite images are taken from the Meteorological&Oceanographic Satellite Data Archival Centre,the organization is responsible for availing satellite cloud images of India and its subcontinent.The proposed cloud image classification shows 94% prediction accuracy with the DCNN framework.