Object detection (OD) in unmanned aerial vehicle (UAV) images faces many challenges, with diverse-scale objects and small being particularly prominent issues. To alleviate these we propose a novel multiscale feature learning fusion network under the guidance of deformable convolution. First, convolution-guided (DCGFL) block is designed backbone to extract more effective features. The DCGFL leverages adaptability convolution shapes scales objects, akin spatial attention. Moreover, it also...

We have developed a scheme that retrieves primary reflections in the two-way traveltime domain by filtering data. The data their own filter removes internal multiple reflections, whereas amplitudes of retrieved are compensated for transmission losses. Application does not require any model information. It consists convolutions and correlations with itself. A truncation time is applied after each convolution or correlation. set can be used as input to construct better velocity than one would...

With the Marchenko method, it is possible to retrieve Green’s functions between virtual sources in subsurface and receivers at surface from reflection data focusing functions. A macro model of needed estimate first arrival; internal multiples are retrieved entirely data. The form input for redatuming by multidimensional deconvolution (MDD). redatumed response free related overburden. Alternatively, can be obtained applying a second function This process called double focusing. It more stable...

We have derived a scheme for retrieving the primary reflections from acoustic surface-reflection response by eliminating free-surface and internal multiple in one step. This does not require model information adaptive subtraction. It consists only of reflection as correlation convolution operator that acts on an intermediate wavefield which we compute capture reflections. For each time instant, keep value source-receiver pair store it new data set. The resulting set contains reflections,...

Internal multiple reflections have been widely considered as coherent noise in measured seismic data, and many approaches developed for their attenuation. The Marchenko elimination (MME) scheme eliminates internal without model information or adaptive subtraction. This was originally derived from coupled equations, but it modified to make independent. It filters primary with two-way traveltimes physical amplitudes data. MME is applied a deepwater field data set the Norwegian North Sea...

The benefits of simultaneous source acquisition are compromised by the challenges dealing with intense blending noise. In this paper, we propose a processing workflow for blended data. incoherent property noise in common-midpoint gathers is utilized applying median filtering along spatial direction after normal moveout (NMO) correction. key step proposed that need to obtain precise velocity estimation which required subsequent NMO Because noise, scan cannot be obtained one step. We can...

Free-surface-related multiples are usually regarded as noise in conventional seismic processing. However, they can provide extra illumination of the subsurface and thus have been used migration procedures, e.g., one- two-way wave-equation migrations. The disadvantage is artifacts generated by crosscorrelation different events, primaries second-order free-surface-related multiples, so effective elimination crucial for multiples. angle domain common image gather (ADCIG) a suitable testing...

Seismic data interpolation plays an indispensable role in common seismic processing workflows. Iterative shrinkage thresholding (IST) and projection onto convex sets (POCS) can both be considered as a specific form of nonlinear shaping regularization. Compared with linear regularization, the version more adaptive because operator is not limited to linear. With combination operator, we introduce faster The new utilizes information previous model better constrain current model. Both synthetic...

We have derived an improved reverse time migration (RTM) scheme to image the medium without artifacts arising from internal multiple reflections. This is based on a revised implementation of Marchenko redatuming using new time-truncation operator. Because truncation operator, we can use time-reversed version standard wavefield-extrapolation operator as initial estimate for retrieving upgoing focusing function. Then, retrieved function be used directly by correlating it with imaging seen...

SUMMARY The transmission compensated primary reflections can be obtained from the single-sided acoustic reflection response in two-way traveltime domain. This is achieved by eliminating free-surface and internal multiple compensating for losses without model information. substantial computational cost of proposed scheme reduced an order magnitude with a fast implementation version. using previously computed filter functions as initial estimate every new truncation time value. We evaluate...

We have compared three data-driven internal multiple reflection elimination schemes derived from the Marchenko equations and inverse scattering series (ISS). The two are similar but use different truncation operators. first scheme creates a new data set without reflections. second does same compensates for transmission losses in primary ISS is equal to result after iteration of Marchenko-based scheme. It can attenuate reflections with residuals. evaluate success these 2D numerical examples....

The Marchenko multiple elimination (MME) and transmission compensation schemes retrieve primary reflections in the two-way traveltime domain without model information or using adaptive subtraction. Both are derived from projected equations similar to each other, but they use different time-domain truncation operators. MME scheme retrieves a new data set internal reflections. transmission-compensated does same additionally compensates for losses can be solved with an iterative algorithm based...

Modeling futures market risk simultaneously influenced by macro low-frequency information and daily factors is a valuable challenge. We propose new general framework for it based on the flexible GARCH-MIDAS model. It uses skewed t distribution to describe asymmetry of long short trading positions, allows different number days per month, can identify optimal combination risky factors. also derive its impact response function how directly influence high-frequency risk. Through an exhaustive...

Spectrum sensing technology is a crucial aspect of modern communication technology, serving as one the essential techniques for efficiently utilizing scarce information resources in tight frequency bands. This paper first introduces three common logical circuit decision criteria hard decisions and analyzes their rigor. Building upon decisions, further method multi-user spectrum based on soft decisions. Then simulates false alarm probability detection curves corresponding to criteria. The...

Seismic images provided by reverse time migration can be contaminated artefacts associated with the of multiples. Multiples corrupt seismic images, producing both false positives, that is focusing energy at unphysical interfaces, and negatives, destructively interfering primaries. Multiple prediction/primary synthesis methods are usually designed to operate on point source gathers therefore computationally demanding when large problems considered. A attractive scheme operates plane-wave...

SUMMARY The Marchenko multiple elimination (MME) scheme is derived from the coupled equations. It proposed for filtering primary reflections with two-way traveltime measured acoustic data. reflection data are used as its own filter and no model information or adaptive subtraction required to apply method. obtained after MME better suited velocity construction artefact-free migration than We a laboratory set evaluate success of results suggest that can be appropriate choice when high-quality...

<div>Federated learning (FL) is an emerging paradigm for decentralized training of machine models on distributed clients, without revealing the data to central server. The scheme may be horizontal, vertical or hybrid (both and horizontal). Most existing research work with deep neural network (DNN) modeling focused horizontal distributions, while schemes are much less studied. In this paper, we propose a generalized algorithm FedEmb, DNN-based learning. idea our characterized by higher...

In acoustic reflector imaging, we deploy sources and receivers outside a volume to collect multisource, multioffset reflection response in order retrieve the internal reflectivity of that volume. It has been shown Green's functions inside can be retrieved by single-sided wavefield focusing acquired data, using so-called functions, which computed solving multidimensional Marchenko equation. Besides this methodology requires background model propagation velocity. We present several imaging...

ABSTRACT Migration, velocity and amplitude analysis are the employed processing steps to find desired subsurface information from seismic reflection data. The presence of free‐surface internal multiples can mask primary reflections for which many methods built. ability separate multiple is desirable. Connecting Marchenko theory with classical one‐dimensional inversion allows understand process elimination as a data‐filtering process. filter fundamental wave field, defined pressure particle...

Multiple reflections are often considered as noise in seismic data. Dealing with them has attracted much attention from industry and academia. A variety of schemes have been developed for removal free-surface multiple success achieved numerical field datasets. However, less directed toward internal reflection removal. The Marchenko elimination scheme is capable to eliminate without model information or adaptive subtraction. This derived coupled Marchenko-type equations by projecting the...

We present a three-dimensional scheme that can be used to compute vertical radar profile from reflection and transmission data measured at two parallel surfaces of dissipative medium. From this double-sided set the response fictitious medium with negative dissipation computed. The computed in construct focusing wavefields. One focuses chosen location subsurface actual other one inside same location. This is then virtual receiver for Green's function. Because up- downgoing parts function are...

<p>Spectrum sensing technology is a crucial aspect of modern communication technology, serving as one the essential techniques for efficiently utilizing scarce information resources in tight frequency bands. This paper first introduces three common logical circuit decision criteria hard decisions and analyzes their rigor. Building upon decisions, further method multi-user spectrum based on soft decisions. Then simulates false alarm probability detection curves corresponding to...