Lossless and near-lossless image compression is of paramount importance to professional users in many technical fields, such as medicine, remote sensing, precision engineering scientific research. But despite rapidly growing research interests learning-based compression, no published method offers both lossless modes. In this paper, we propose a unified powerful deep lossy plus residual (DLPR) coding framework for compression. the mode, DLPR system first performs then residuals. We solve...

This paper presents an energy ratio-based method and a wavelet-based cascaded adaptive filter (CAF) for detecting removing baseline drift from pulse waveforms. Experiments on 50 simulated five hundred real signals demonstrate that this CAF outperforms traditional filters both in preserving the diagnostic information of

Transformer-based entropy models have gained prominence in recent years due to their superior ability capture long-range dependencies probability distribution estimation compared convolution-based methods. However, previous transformer-based suffer from a sluggish coding process pixel-wise autoregression or duplicated computation during inference. In this paper, we propose novel model called GroupedMixer, which enjoys both faster speed and better compression performance than Specifically,...

Learned lossless image compression has achieved significant advancements in recent years. However, existing methods often rely on training amortized generative models massive datasets, resulting sub-optimal probability distribution estimation for specific testing images during encoding process. To address this challenge, we explore the connection between Minimum Description Length (MDL) principle and Parameter-Efficient Transfer Learning (PETL), leading to development of a novel...

With the development of society and advancement technology, intelligent robots have been widely used in various fields. At same time, Simultaneous Localization Mapping (SLAM) technology is a key research field robots. However, dynamic environments, achieving accurate robust visual SLAM remains major challenge. In this paper, we propose method based on improved YOLOv8 fused with ORB-SLAM3 to address dense point cloud environments. Our proposed successfully integrates real-time object...

In this work, the state-of-the-art infrared variable angle spectroscopic ellipsometry (IR-VASE) and first-principles molecular dynamics (FPMD) method were combined to obtain dielectric functions of MgO crystal in spectral range 300–1000 cm−1 for temperatures up 1950 K. The IR-VASE can measure at ranging from 300 573 K reproduce previous infrared-reflectivity experiments. As temperature increases, it demonstrates that amplitude dominant absorption peak centered around 400 reduces, width...

This paper presents an automatic 3D ear reconstruction method based on binocular stereo vision. At first, we calibrate the vision system by Zhang's method. Then quasi-dense matching is performed. We use SIFT feature approach and coarse to fine strategy compute seed matches. The adapted match propagation algorithm with known epipolar geometry constraint used for obtain correspondence points. Finally model can be reconstructed triangulation results are optimized using bundle adjustment...

Speeded Up Robust Features (SURF) algorithm is a fast robust local feature matching method. Because of the advantage invariance scale and rotation, SURF has been widely used in image registration. However, due to acquired angle, imaging mode, different resolution, becomes quite difficult This paper presents normalized which can reduce impact hue difference between remote sensing images. Combined with idea two-way matching, able improve accuracy The experimental results show keep greater...

In certain operational radar modes, slow ground moving targets are detected over several processing intervals using space-time adaptive processing. This enables use of Bayesian filtering and smoothing algorithms for estimation time-varying target parameters. this paper, some investigated. The Cram´er-Rao bounds based on subsets measurements (range, angle Doppler) derived typical maneuvering compared against simulated results from filters. performance is also evaluated real data obtained DRDC...

The multi-model adaptive hull deformation estimation algorithm is proposed for the uncertainty of system parameters and statistical characteristics measurement noise. Based on idea, a variable parameter designed. By pre-estimating main frequency, problem frequency in actual solved. Sage-Husa filtering volume Kalman filter designed to track estimate noise covariance matrix real time, which effectively solves effectiveness verified by simulation. When are unknown, accuracy angle better than...

In this paper, we introduce an improved version of DAISY descriptor algorithm for fast and high-quality image key point matching. Since has many prominent advantages but lacks the ability handling large in-plane rotation, thus construct our Rotation-Invariant to effectively settle shortcoming. We first extract points images by Harris corner detector, then use histogram data every assign a local orientation each. During step constructing descriptors, rotate coordinate axis its orientation,...

Recent advances in learning-based methods have markedly enhanced the capabilities of image compression. However, these struggle with high bit-depth volumetric medical images, facing issues such as degraded performance, increased memory demand, and reduced processing speed. To address challenges, this paper presents Bit-Division based Lossless Volumetric Image Compression (BD-LVIC) framework, which is tailored for volume The BD-LVIC framework skillfully divides into two lower segments: Most...

Learned lossless image compression has achieved significant advancements in recent years. However, existing methods often rely on training amortized generative models massive datasets, resulting sub-optimal probability distribution estimation for specific testing images during encoding process. To address this challenge, we explore the connection between Minimum Description Length (MDL) principle and Parameter-Efficient Transfer Learning (PETL), leading to development of a novel...

This paper applies one kind of methods automatic image registration, Scale Invariant Feature Transform(SIFT), into the region remote sensing images. It has been found that SIFT can extract features invariant to scale and rotation with specific key point descriptors as matching points for registration Experimental results are discussed in last part this paper, confirmed validity proposed method.

Aiming at the problem that moving target on ground/sea surface is difficult to detect by traditional multi-UAV cooperative parallel search strategy, a strategy with formation shape of V proposed in this paper. This method improves probability detecting reducing coverage blind area airborne sensor. Firstly, described detail. Then, motion model UAV, for sensor detection, and discriminant detection are respectively established. Finally, simulation carried out Matlab/M-file programming. The...

To improves tracking drift which often occurs in adaptive tracking, an algorithm based on the fusion of and detection is proposed this paper. Firstly, object frame by via color histogram particle filtering. Secondly, reversely validating result Finally, relocating SIFT features matching voting when occurs. Object appearance model updated at same time. The can not only sense but also relocate whenever needed. Experimental results demonstrate that outperforms state-of-the-art algorithms many...