The propose of this paper is to design the circularly polarized antenna for mobile base station, using a curved strip dipole associated with U-shaped reflector and multi-layer metallic rod elec-tromagnetic band gap (E...The propose of this paper is to design the circularly polarized antenna for mobile base station, using a curved strip dipole associated with U-shaped reflector and multi-layer metallic rod elec-tromagnetic band gap (EBG). It can be used to simplify the single feed system of an antenna. The advantages of this proposed antenna are easy fabrication and installation, high gain and light weight. Moreover, it provides a sectoral radiation pattern, a main beam having a narrow beam width in the vertical direction and wider beamwidth in the horizontal direction, which are appro-priate for mobile phone base station. The half-power beam in the H-plane and E-plane are 60 and 14.4 degrees, respectively. The antenna is designed and analyzed by using a computer simulation technology (CST). In addition, S11, axial ratio, radiation pattern and gain are displayed. The designed technique could be confirmed by a measurement resulting from prototype antenna corresponding to simulation results. The proposed antenna has a bandwidth covering the frequency range of 1870 - 2170 MHz, and the gain of the antenna increases up to 15.11 dBi.展开更多
近年来,随着深度学习的发展,在自然街景下的文本检测取得了巨大的进步,但在多方向和弯曲文本及对比度低的文本检测中的效果仍不理想。因此,针对弯曲文本和对比度低的文本的检测问题,提出了一种融合多尺度模块的文本检测方法,并通过检测...近年来,随着深度学习的发展,在自然街景下的文本检测取得了巨大的进步,但在多方向和弯曲文本及对比度低的文本检测中的效果仍不理想。因此,针对弯曲文本和对比度低的文本的检测问题,提出了一种融合多尺度模块的文本检测方法,并通过检测效果的提升,提高端到端文本识别的识别效果。针对RFB(Receptive Field Block)模块在下采样后局部信息丢失的问题,在RFB模块中嵌入极化自注意力(Polarized Self-Attention)机制以改进RFB来提取有效文本特征,提高特征图表征效果。针对特征金字塔(FPN)提取的特征不足、感受野小的问题,将改进的RFB模块嵌入特征金字塔(FPN)模块以增强特征提取融合。针对特征分布不确定性及远距离特征融合效果不佳的问题,引入条形池化(Strip Pooling)模块,进而提升检测方法的鲁棒性。在公开数据集Total-Text上的实验结果表明,该算法的F-measure值在端到端文本识别没有词汇表的情形下与目前高效的MaskTextSpotterV3相比高了0.3百分点,而在有词汇表的情形下则高出了0.2百分点;而在仅文本检测的情形下,该方法也有较为良好的表现。展开更多
文摘The propose of this paper is to design the circularly polarized antenna for mobile base station, using a curved strip dipole associated with U-shaped reflector and multi-layer metallic rod elec-tromagnetic band gap (EBG). It can be used to simplify the single feed system of an antenna. The advantages of this proposed antenna are easy fabrication and installation, high gain and light weight. Moreover, it provides a sectoral radiation pattern, a main beam having a narrow beam width in the vertical direction and wider beamwidth in the horizontal direction, which are appro-priate for mobile phone base station. The half-power beam in the H-plane and E-plane are 60 and 14.4 degrees, respectively. The antenna is designed and analyzed by using a computer simulation technology (CST). In addition, S11, axial ratio, radiation pattern and gain are displayed. The designed technique could be confirmed by a measurement resulting from prototype antenna corresponding to simulation results. The proposed antenna has a bandwidth covering the frequency range of 1870 - 2170 MHz, and the gain of the antenna increases up to 15.11 dBi.
文摘近年来,随着深度学习的发展,在自然街景下的文本检测取得了巨大的进步,但在多方向和弯曲文本及对比度低的文本检测中的效果仍不理想。因此,针对弯曲文本和对比度低的文本的检测问题,提出了一种融合多尺度模块的文本检测方法,并通过检测效果的提升,提高端到端文本识别的识别效果。针对RFB(Receptive Field Block)模块在下采样后局部信息丢失的问题,在RFB模块中嵌入极化自注意力(Polarized Self-Attention)机制以改进RFB来提取有效文本特征,提高特征图表征效果。针对特征金字塔(FPN)提取的特征不足、感受野小的问题,将改进的RFB模块嵌入特征金字塔(FPN)模块以增强特征提取融合。针对特征分布不确定性及远距离特征融合效果不佳的问题,引入条形池化(Strip Pooling)模块,进而提升检测方法的鲁棒性。在公开数据集Total-Text上的实验结果表明,该算法的F-measure值在端到端文本识别没有词汇表的情形下与目前高效的MaskTextSpotterV3相比高了0.3百分点,而在有词汇表的情形下则高出了0.2百分点;而在仅文本检测的情形下,该方法也有较为良好的表现。
