Building Footprint Identification Using Remotely Sensed Images: A Compressed Sensing-Based Approach to Support Map Updating

Geomatics, Journal Year: 2025, Volume and Issue: 5(1), P. 7 - 7

Published: Jan. 31, 2025

Semantic segmentation of remotely sensed images for building footprint recognition has been extensively researched, and several supervised unsupervised approaches have presented adopted. The capacity to do real-time mapping precise on a significant scale while considering the intrinsic diversity urban landscape in data consequences. This study presents novel approach delineating footprints by utilizing compressed sensing radial basis function technique. At feature extraction stage, small set random features built-up areas is extracted from local image windows. are used train neural network perform classification; thus, learning classification carried out domain. By virtue its ability represent characteristics reduced dimensional space, scheme shows promise being robust face variability inherent images. Through comparison proposed method with numerous state-of-the-art different spatial resolutions clutter, we establish robustness prove viability. Accuracy assessment performed segmented footprints, comparative analysis terms intersection over union, overall accuracy, precision, recall, F1 score. achieved scores 93% 90.4% 91.1% score, even when dealing drastically features. results demonstrate that methodology yields substantial enhancements accuracy decreases dimensionality.

Language: Английский

---

TLBP: Tomography‐Aided Local Binary Patterns With High Discrimination for Image Classification

IET Image Processing, Journal Year: 2025, Volume and Issue: 19(1)

Published: Jan. 1, 2025

ABSTRACT Local binary patterns (LBP) play a vital role in image classification as computationally efficient feature descriptor. A crucial reason for its limitation of discriminability is the lack neighbourhood information description from global perspective. Previous research has attempted to improve performance by introducing thresholds, but such threshold selection not optimal. To address this issue, we propose novel tomography‐aided local (TLBP), inspired tomographic process sample separation. TLBP considers constructing visual representations under multi‐level non‐local compensate LBP possessing only single shallow feature. In addition basic features context, captures refined greyscale through multi‐quantile thresholds perspective, thereby greatly enhancing discriminability. Experimental results texture classification, face recognition, and hyperspectral pixel‐wise demonstrate that proposed descriptor outperforms competitors, achieving 94.39% (KTH‐TIPS), 81.22% (KTH‐TIPS‐ROT), 93.81% (Indian Pines), 99.85% (Salinas), 99.50% (ORL) accuracy. Furthermore, T‐variants apply idea classic descriptors significantly, especially their rotation‐invariant versions.

Language: Английский

---

Efficient detection and classification of brain tumors using a novel local binary pattern

Academia Engineering, Journal Year: 2025, Volume and Issue: 2(1)

Published: March 14, 2025

Local binary pattern (LBP) is a computationally inexpensive feature descriptor popularly used for detecting and classifying images. However, the directional attributes associated with conventional LBP lead to generation of undesirable shades noisy textures. Consequently, it impairs features corresponding edges intensity gradients in an image. To address this problem, article, we have proposed novel LBP-Like (LBP-L) transform, which could be as efficient alternative LBP. The LBP-L image does not change due horizontal vertical flipping images or when rotated through 90°, 180°, 270°. Besides, maps uniform zones same space irrespective orientation. It has significant potential boost detection classification brain tumors magnetic resonance (MR) performance been tested two benchmark datasets having 3064 7023 MR three different types, namely, meningioma, glioma, pituitary, along without tumors. accuracies on Kaggle Figshare are higher than those its variants reported literature. shown that computational complexity comparable traditional Moreover, found lower computation time compared well-known variants. Therefore, more substitute latter.

Language: Английский

---

ELMP-Net: The successive application of a randomized local transform for texture classification

Pattern Recognition, Journal Year: 2024, Volume and Issue: 153, P. 110499 - 110499

Published: April 14, 2024

Language: Английский

---

The Prediction and Evaluation of Surface Quality during the Milling of Blade-Root Grooves Based on a Long Short-Term Memory Network and Signal Fusion

Sensors, Journal Year: 2024, Volume and Issue: 24(15), P. 5055 - 5055

Published: Aug. 5, 2024

The surface quality of milled blade-root grooves in industrial turbine blades significantly influences their mechanical properties. texture reveals the interaction between tool and workpiece during machining process, which plays a key role determining quality. In addition, there is significant correlation acoustic vibration signals features. However, current research on still relatively limited, most considers only single signal. this paper, 160 sets field data were collected by multiple sensors to study groove. A feature prediction method based signal fusion proposed evaluate Fast Fourier transform (FFT) used process signal, clean smooth features are extracted combining wavelet denoising multivariate smoothing denoising. At same time, gray-level co-occurrence matrix, image different angles groove describe fused input, output establish model. After predicting features, evaluated setting threshold value. selected all sample data, final judgment accuracy 90%.

Language: Английский

---