Background estimation is an efficient infrared (IR) small target detection method. However, to deal with unknown targets, the estimation window in existing algorithms should be adjusted to perform multiscale detection, and requires a lot of calculations. Besides, the stages during and after estimation have received wide attention in existing algorithms, but the research on the stages before estimation is insufficient. Moreover, existing algorithms typically regard the maximum value of different orientations as the estimation value. However, when a dim target is adjacent to high brightness background, it is easily submerged. This paper proposes a three-layer estimation window to detect targets of different sizes with only a single-scale calculation. The enhanced closest-mean background estimation method is then proposed and carefully designed before, during and after the estimation. Before estimation, the matched filter is adopted to improve the image signal-to-noise ratio. During estimation, the principle of closest-mean is proposed to suppress high brightness background. After estimation, a ratio-difference operation is performed to enhance the true target and suppress the background simultaneously. A simple checking mechanism is proposed to further improve the detection performance. Experiments on some IR images demonstrate the effectiveness and robustness of the proposed method. Compared with existing algorithms, the proposed method has better target enhancement, background suppression, and computational efficiency.

Local contrast methods are efficient approaches for infrared (IR) small target detection. However, since its baseline is obtained locally, a local contrast method can work well only when the target is locally prominent, which is not always the case when the background is very complex. In this paper, a new detection framework named true global contrast measure (TGCM) is proposed, in which its baseline is obtained by global decomposition, so that it can handle situations where the target is not locally salient. For example, when it is adjacent to some bright background. The complex edges / corners and noises that are easily lost during decomposition will be suppressed through some simple but ingenious design: A Gaussian filtering operation is adopted to suppress noises, and a local max dilation is utilized on the separated baseline to maintain as much edge / corner information as possible. The residual information during decomposition will be used as a weighting function to further suppress clutters. Experiments on both real and simulated IR images demonstrated the effectiveness of the proposed method. Compared to some existing algorithms, it has advantages in target enhancement and background suppression. Especially, it can achieve good performance in certain scenarios where the target is not locally prominent.

Long-time integration technique is an effective way of improving target detection performance for unmanned aerial vehicle (UAV) in the passive bistatic radar (PBR) field, while range migration (RM) and Doppler frequency migration (DFM) may have a major effect on the integration capability as a result of target maneuverability. This letter proposed an innovative long-time coherent integration approach, regarded as Continuous Radon-matched filtering process (CRMFP), for low-observable UAV target in PBR. It not only mitigates the RM by collaborative research in range and velocity dimensions but also compensates the DFM and ensures the coherent integration through the matched filtering process (MFP). Real-life data following detailed analysis verify that the proposed method can overcome the Doppler mismatch influence and acquire comparable detection performance.

Chinese Medical Named Entity Recognition (CMNER) seeks to identify and extract medical entities from unstructured medical texts. Existing methods often depend on single-modality representations and fail to fully exploit the complementary nature of different features. This paper presents a multimodal information fusion-based approach for medical named entity recognition, integrating a hybrid attention mechanism. A Dual-Stream Network architecture is employed to extract multimodal features at both the character and word levels, followed by deep fusion to enhance the model’s ability to recognize medical entities. The Cross-Stream Attention mechanism is introduced to facilitate information exchange between different modalities and capture cross-modal global dependencies. Multi-Head Attention is employed to further enhance feature representation and improve the model’s ability to delineate medical entity boundaries. The Conditional Random Field (CRF) layer is used for decoding, ensuring global consistency in entity predictions and thereby enhancing recognition accuracy and robustness. The proposed method achieves F1 scores of 65.26%, 80.31%, and 86.73% on the CMeEE-V2, IMCS-V2-NER, and CHIP-STS datasets, respectively, outperforming other models and demonstrating significant improvements in medical entity recognition accuracy and multiple evaluation metrics.

This study examines the target detection problem in passive bistatic radar. It is well-known that traditional target detection methods are designed under the assumption that the reference signal is ideally free of target echoes. However, this is problematic for passive bistatic radar as the reference channel has often a large main-lobe width and high side-lobe level. As will be shown in this paper, the target echoes contained in the reference channel degrade the target signal to clutter ratio (SCR) and increase the false alarm rate. Therefore, a matched rule for making a distinction between the actual targets and their ghost targets is established, and a clean method for eliminating the redundant target echoes from the reference channel is proposed. Based on simulations it has been shown that as the target echoes are removed from the reference channel, the ghost targets corresponding to this effect disappear after cross-correlation with the surveillance channel.

This paper studies the girth of Tanner (5,13)-regular QC-LDPC codes. We present the conditions for the existence of cycles of lengths less than 12 in Tanner (5,13)-regular QC-LDPC codes of length ${13p}$ where ${p}$ is a prime of the form ${65i+1}$ , and transfer these conditions into the equations in a 65th root of unity of the prime field ${\mathbb{F}_p}$ . By checking the existence of solutions for these equations over ${\mathbb{F}_p}$ , we derive the girth distribution of Tanner (5,13)-regualr QC-LDPC codes.