A Scene Semantic Recognition Method
of Remote Sensing Images Based
on CSIFT and PLSA
Yan-li Sun, Wei Zhou
(&)
, Ya-zhou Zhang, and Jialong Jin
Naval Aviation University, Yantai 264001, China
yeaweam@gmail.com
Abstract. Aiming at the fast recognition of local image scenes with clear
semantics in high-resolution remote sensing images, such as ports, airports and
oil depots, a visual feature representation method for remote sensing images
based on CSIFT features and a scene semantic recognition method based on
PLSA is proposed. Experiments on typical remote sensing image scenes fully
verify the effectiveness of proposed method.
Keywords: Semantic recognition of scenes Á Visual feature Á PLSA
1 Introduction
Information interpretation of high-resolution remote sensing images usually starts with
the detection and recognition of specific types of objects [1]. Cognitive understanding
of the scene can only be achieved after detecting and recognizing all kinds of targets in
the scene. However, in the military reconnaissance, we often need to quickly identify
local image scenes with definitive semantic meaning such as ports, airports, oil depot
from high-resolution remote sensing images. These local scenes often correspond to a
complex target group, contain many different types of rigid sub-targets according to
certain spatial topological relations [2]. If by conventional method, detect subgoals of
complex target group firstly, then to identify the complex target group of semantic
properties by reasoning the spatial topological relations, but the processing efficiency is
very low. And the analysis of sub-target is mainly on pixel level by using its implicit
heuristic features. There is a significant gap between this complete pixel-level processing and image semantic understanding. It is very important to establish a semantic
recognition method to classify complex local scene in remote sensing image quickly.
And the scene semantic recognition result will be useful for the detection and recognition of specific targets in large remote sensing images. In this paper, related concepts
and schemes of image scene recognition are first summarized. Then visual feature
representation method for remote sensing image is provided and also a semantic
recognition algorithm based on PLSA is given. Finally it is verified by experiments in
10 typical remote sensing image scenes.
© Springer Nature Singapore Pte Ltd. 2020
Q. Liang et al. (Eds.): Artificial Intelligence in China, LNEE 572, pp. 378–384, 2020.
https://doi.org/10.1007/978-981-15-0187-6_44
of Remote Sensing Images Based
on CSIFT and PLSA
Yan-li Sun, Wei Zhou
(&)
, Ya-zhou Zhang, and Jialong Jin
Naval Aviation University, Yantai 264001, China
yeaweam@gmail.com
Abstract. Aiming at the fast recognition of local image scenes with clear
semantics in high-resolution remote sensing images, such as ports, airports and
oil depots, a visual feature representation method for remote sensing images
based on CSIFT features and a scene semantic recognition method based on
PLSA is proposed. Experiments on typical remote sensing image scenes fully
verify the effectiveness of proposed method.
Keywords: Semantic recognition of scenes Á Visual feature Á PLSA
1 Introduction
Information interpretation of high-resolution remote sensing images usually starts with
the detection and recognition of specific types of objects [1]. Cognitive understanding
of the scene can only be achieved after detecting and recognizing all kinds of targets in
the scene. However, in the military reconnaissance, we often need to quickly identify
local image scenes with definitive semantic meaning such as ports, airports, oil depot
from high-resolution remote sensing images. These local scenes often correspond to a
complex target group, contain many different types of rigid sub-targets according to
certain spatial topological relations [2]. If by conventional method, detect subgoals of
complex target group firstly, then to identify the complex target group of semantic
properties by reasoning the spatial topological relations, but the processing efficiency is
very low. And the analysis of sub-target is mainly on pixel level by using its implicit
heuristic features. There is a significant gap between this complete pixel-level processing and image semantic understanding. It is very important to establish a semantic
recognition method to classify complex local scene in remote sensing image quickly.
And the scene semantic recognition result will be useful for the detection and recognition of specific targets in large remote sensing images. In this paper, related concepts
and schemes of image scene recognition are first summarized. Then visual feature
representation method for remote sensing image is provided and also a semantic
recognition algorithm based on PLSA is given. Finally it is verified by experiments in
10 typical remote sensing image scenes.
© Springer Nature Singapore Pte Ltd. 2020
Q. Liang et al. (Eds.): Artificial Intelligence in China, LNEE 572, pp. 378–384, 2020.
https://doi.org/10.1007/978-981-15-0187-6_44