The benefits of simultaneous source acquisition are compromised by the challenges of dealing with intense blending noise. In this paper, we propose a processing workflow for blended data. The incoherent property of blending noise in the common-midpoint gathers is utilized for applying median filtering along the spatial direction after normal moveout (NMO) correction. The key step in the proposed workflow is that we need to obtain a precise velocity estimation which is required by the subsequent NMO correction. Because of the intense blending noise, the velocity scan cannot be obtained in one step. We can recursively polish both the deblended result and the velocity estimation by deblending using the updated velocity estimation and velocity scanning using the updated deblended result. We use synthetic and field data examples to demonstrate the performance of the proposed approach. The migrated image of deblended data is cleaner than that of blended data and is similar to that of unblended data.
One of the most crucial estimates retrieved from measured seismic reflection data is the subsurface image. The image provides detailed information of the subsurface of the Earth. Seismic reflection data consists of so-called primary and multiple reflections. Primary reflections are events that have been reflected a single time, while multiple reflections have been reflected multiple times before they are recorded by the receivers. Most current migration algorithms assume all reflections in the data are primary reflections. Hence, in order to retrieve an accurate image of the subsurface, multiple reflections need to be eliminated before migration. Keeping the multiple reflections in the measured seismic reflection data will lead to a sub-optimal image of the subsurface, because the multiple reflections will be imaged as if they were primary reflections. Such artefacts in the image can cause erroneous interpretation...
A modified implementation of Marchenko redatuming leads to a filter that removes internal multiples from reflection data. It produces local reflectivity at two-way travel time. The method creates new primary reflections resulting from emitted events that eliminate internal multiples. We call these non-physical primaries and their presence is a disadvantage. The advantage is that the filter is model free. We give the 3D filter and demonstrate with 1D arguments that starting the focusing wavefield with a unit impulse at zero time, while focusing below the bottom reflector, is the choice that leads to a model free implementation. The starting impulse generates the reflection data. Every later emitted pulse eliminates an internal multiple somewhere in the model and helps removing the transmission amplitude effects in a physical primary. We show that the amplitude of the non-physical primaries are a product of three reflections, making them generally smaller than those of the physical primaries. A 2D modeled shotgather at different stages of filtering the data shows that the filter works well. Presentation Date: Wednesday, September 27, 2017 Start Time: 4:45 PM Location: 370A Presentation Type: ORAL
We have developed a scheme that retrieves primary reflections in the two-way traveltime domain by filtering the data. The data have their own filter that removes internal multiple reflections, whereas the amplitudes of the retrieved primary reflections are compensated for two-way transmission losses. Application of the filter does not require any model information. It consists of convolutions and correlations of the data with itself. A truncation in the time domain is applied after each convolution or correlation. The retrieved data set can be used as the input to construct a better velocity model than the one that would be obtained by working directly with the original data and to construct an enhanced subsurface image. Two 2D numerical examples indicate the effectiveness of the method. We have studied bandwidth limitations by analyzing the effects of a thin layer. The presence of refracted and scattered waves is a known limitation of the method, and we studied it as well. Our analysis indicates that a thin layer is treated as a more complicated reflector, and internal multiple reflections related to the thin layer are properly removed. We found that the presence of refracted and scattered waves generates artifacts in the retrieved data.
We present a one-dimensional lossless scheme to compute an image of a dissipative medium from two single-sided reflection responses. One reflection response is measured at or above the top reflector of a dissipative medium and the other reflection response is computed as if measured at or above the top reflector of a medium with negative dissipation which we call the effectual medium. These two reflection responses together can be used to construct the approximate reflection data of the corresponding lossless medium by multiplying and taking the square root in time domain. The corresponding lossless medium has the same reflectors as the dissipative medium. Then the constructed reflection data can be used to compute the focusing wavefield which focuses at the chosen location in subsurface of the dissipative medium. From the focusing function and constructed reflection response the Green's function for a virtual receiver can be obtained. Because the up- and downgoing parts of the Green's function are retrieved separately, these are used to compute the image. We show with an example that the method works well for a sample in a synthesized waveguide that could be used for measurements in a laboratory. Presentation Date: Wednesday, October 19, 2016 Start Time: 10:20:00 AM Location: Lobby D/C Presentation Type: POSTER
In acoustic reflector imaging, we deploy sources and receivers outside a volume to collect a multisource, multioffset reflection response in order to retrieve the internal reflectivity of that volume. It has been shown that Green's functions inside the volume can be retrieved by single-sided wavefield focusing of the acquired reflection data, using so-called focusing functions, which can be computed by solving a multidimensional Marchenko equation. Besides the reflection data, this methodology requires a background model of the propagation velocity. We present several imaging conditions to retrieve the internal reflectivity of an acoustic medium with correct amplitudes and without artifacts, using the Green's functions and focusing functions that are derived from the Marchenko equation. We distinguish three types of imaging: 1) imaging by deconvolution, 2) imaging by double focusing, and 3) imaging by cross correlation. In all cases, reflectors can be approached either from above or from below. Imaging by deconvolution or double focusing requires single-sided illumination (meaning that sources and receivers are deployed at a single boundary above the volume only), whereas imaging by cross correlation requires double-sided illumination (meaning that sources and receivers are placed at two boundaries enclosing the volume). In order to achieve double-sided illumination, the required reflection response at the lower boundary can either be physically recorded or it can be retrieved from the reflection response at the upper boundary. When imaging by deconvolution or double focusing, the internal reflectivity is retrieved solely from primary reflections. When imaging by cross correlation, multiple reflections are focused at the image points, such that they contribute physically to the retrieved reflectivity values. This special feature can be beneficial for imaging weakly illuminated sections of strongly heterogeneous media.
We have compared three data-driven internal multiple reflection elimination schemes derived from the Marchenko equations and inverse scattering series (ISS). The two schemes derived from Marchenko equations are similar but use different truncation operators. The first scheme creates a new data set without internal multiple reflections. The second scheme does the same and compensates for transmission losses in the primary reflections. The scheme derived from ISS is equal to the result after the first iteration of the first Marchenko-based scheme. It can attenuate internal multiple reflections with residuals. We evaluate the success of these schemes with 2D numerical examples. It is shown that Marchenko-based data-driven schemes are relatively more robust for internal multiple reflection elimination at a higher computational cost.
Object detection (OD) in unmanned aerial vehicle (UAV) images faces many challenges, with diverse-scale objects and small objects being particularly prominent issues. To alleviate these challenges, we propose a novel multiscale feature learning and feature fusion network under the guidance of deformable convolution. First, a deformable convolution-guided feature learning (DCGFL) block is designed in the backbone to extract more effective multiscale features. The DCGFL block leverages the adaptability of deformable convolution to the shapes and scales of objects, akin to spatial attention. Moreover, it also employs channel attention to identify important feature maps. Hence, the proposed backbone possesses the functionality of spatial attention and channel attention. Second, in the neck, we devise a simple generalized feature pyramid network (SimpleGFPN) with several deformable convolution-guided feature fusion (DCGFF) blocks to fuse multiscale features. The proposed neck has cross-layer and cross-scale pathways, facilitating effective information exchange and fusion between shallow spatial and deep semantic features. Third, the SIoU loss is used to better model the bounding box regression loss. Finally, experimental results on the VisDrone2021 and UAVDT datasets show that the proposed method outperforms the compared OD methods. In terms of mean average precision, we obtain 37.8% on VisDrone2021 and 18.5% on UAVDT.
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
