Satellite image retrieval of random forest (rf-PNN) based probablistic neural network

• Published in: Earth science informatics Vol. 15; no. 2; pp. 941 - 949
• Main Author: Bharatha Devi, N.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.06.2022; Springer Nature B.V
• Subjects: Classification; Datasets; Deep learning; Discrete cosine transform; Earth and Environmental Science; Earth Sciences; Earth System Sciences; Feature extraction; Image management; Image retrieval; Image segmentation; Information Systems Applications (incl.Internet); Neural networks; Ontology; Research Article; Satellite imagery; Semantics; Simulation and Modeling; Space Exploration and Astronautics; Space Sciences (including Extraterrestrial Physics; rf-PNN; Linear Binary Pattern; Adaptive median filter; Discrete Cosine Stockwell Transform (DCST); Discrete Cosine Transform based Discrete Orthogonal Stockwell Transform (DCT-DOST)
• ISSN: 1865-0473; 1865-0481
• DOI: 10.1007/s12145-021-00759-3

The image retrieval process grows a massive problem due to the huge number of data exist on the web. So, the accurate retrieval of satellite image of the given query is one of the necessary requirement. The paper propose a classifier of Probabilistic Neural Network based Random Forest (rf-PNN), which is retrieving an exact match of a classified data as per user’s need. Various techniques of Adaptive Median Filter (for pre-processing), Discrete Cosine Transform based Discrete Orthogonal Stockwell Transform (for segmentation) and Linear Binary Pattern (for feature extraction) are presented to process the trained dataset as well as given query. Then, both the feature extracted samples are assigned to compare with the classified network. The experimental setup is demonstrated on MATLAB tool. Then the relevant feature retrieval are analyze under the performance measures of 92% accurate rate, sensitivity for 89.25%, specificity for 94.1% and precision at a rate of 90.08%.