The representation and similarity measurement of geospatial scenes are fundamental tasks in spatial cognition, with broad applications in spatial information retrieval, scene matching, and multi-source data fusion. With the advancement of GeoAI, the adoption of deep learning methods has significantly improved the precision and efficiency of similarity measurement. This paper provides a systematic review of recent advances in geospatial scene similarity measurement. First, it summarizes traditional geospatial scene representation methods and categorizes conventional similarity models into three types: the gradual transformation model based on conceptual neighborhood, the multi-factor weight assignment model, and the association graph structure model. Subsequently, the paper focuses on deep learning-based methods, analyzing raster, vector, and multimodal representations based on CNN, GNN, and Transformer architectures. Furthermore, similarity measurement techniques are classified into three learning paradigms: supervised metric learning, unsupervised representation learning, and self-supervised contrastive learning, along with a review of representative studies and application scenarios. Finally, the paper analyzes key challenges in the current research and looks forward to future development directions. It provides theoretical support and methodological references to advance the theory of geospatial similarity measurement and the construction of intelligent spatial cognitive systems.

Deep-learning-based salient object detection (SOD) has achieved significant success in recent years. The SOD focuses on the context modeling of the scene information, and how to effectively model the context relationship in the scene is the key. However, it is difficult to build an effective context structure and model it. In this article, we propose a novel SOD method called dynamic and adaptive graph convolutional network (DAGCN) that is composed of two parts, adaptive neighborhood-wise graph convolutional network (AnwGCN) and spatially restricted K-nearest neighbors (SRKNN). The AnwGCN is novel adaptive neighborhood-wise graph convolution, which is used to model and analyze the saliency context. The SRKNN constructs the topological relationship of the saliency context by measuring the non-Euclidean spatial distance within a limited range. The proposed method constructs the context relationship as a topological graph by measuring the distance of the features in the non-Euclidean space, and conducts comparative modeling of context information through AnwGCN. The model has the ability to learn the metrics from features and can adapt to the hidden space distribution of the data. The description of the feature relationship is more accurate. Through the convolutional kernel adapted to the neighborhood, the model obtains the structure learning ability. Therefore, the graph convolution process can adapt to different graph data. Experimental results demonstrate that our solution achieves satisfactory performance on six widely used datasets and can also effectively detect camouflaged objects. Our code will be available at: https://github.com/CSIM-LUT/DAGCN.git.

This paper presents an algebraic vector quantized codebook excited linear prediction (AVQ-CELP) speech codec. The objective is to enhance the half rate mode of IS-127, the enhanced variable rate codec (EVRC). In AVQ-CELP scheme, only the perceptually important components are encoded, and the selection of the components is done in a way similar to the ACELP. An open-loop procedure is used to select the subvectors. The selected sub-vectors are concatenated and vector quantized. An analysis-by-synthesis strategy is used to determine the optimal excitation. The generalized Lloyd algorithm (GLA) is used to optimize the AVQ codebook. In order to improve the synthesis quality of voiced frames, a two-pulse version of ACELP is used in the strong voiced frames. The proposed algorithm was incorporated in the Nokia CDMA handset prototype. Under a joint collaboration effort with SK Telecom, a field-testing was performed in Korea to evaluate the performance of the proposed AVQ algorithm. The results indicate a considerable improvement relative to the standard EVRC operating at the maximum half-rate.