scholarly journals SC-SM CAM: An Efficient Visual Interpretation of CNN for SAR Images Target Recognition

2021 ◽  
Vol 13 (20) ◽  
pp. 4139
Author(s):  
Zhenpeng Feng ◽  
Hongbing Ji ◽  
Ljubiša Stanković ◽  
Jingyuan Fan ◽  
Mingzhe Zhu

Convolutional neural networks (CNNs) have successfully achieved high accuracy in synthetic aperture radar (SAR) target recognition; however, the intransparency of CNNs is still a limiting or even disqualifying factor. Therefore, visually interpreting CNNs with SAR images has recently drawn increasing attention. Various class activation mapping (CAM) methods are adopted to discern the relationship between CNN’s decision and image regions. Unfortunately, most existing CAM methods are based on optical images; thus, they usually lead to a limiting visualization effect for SAR images. Although a recently proposed Self-Matching CAM can obtain a satisfactory effect for SAR images, it is quite time-consuming, due to there being hundreds of self-matching operations per image. G-SM-CAM reduces the time of such operation dramatically, but at the cost of visualization effect. Based on the limitations of the above methods, we propose an efficient method, Spectral-Clustering Self-Matching CAM (SC-SM CAM). Spectral clustering is first adopted to divide feature maps into groups for efficient computation. In each group, similar feature maps are merged into an enhanced feature map with more concentrated energy in a specific region; thus, the saliency heatmaps may more accurately tally with the target. Experimental results demonstrate that SC-SM CAM outperforms other SOTA CAM methods in both effect and efficiency.

2021 ◽  
Vol 13 (9) ◽  
pp. 1772
Author(s):  
Zhenpeng Feng ◽  
Mingzhe Zhu ◽  
Ljubiša Stanković ◽  
Hongbing Ji

Synthetic aperture radar (SAR) image interpretation has long been an important but challenging task in SAR imaging processing. Generally, SAR image interpretation comprises complex procedures including filtering, feature extraction, image segmentation, and target recognition, which greatly reduce the efficiency of data processing. In an era of deep learning, numerous automatic target recognition methods have been proposed based on convolutional neural networks (CNNs) due to their strong capabilities for data abstraction and mining. In contrast to general methods, CNNs own an end-to-end structure where complex data preprocessing is not needed, thus the efficiency can be improved dramatically once a CNN is well trained. However, the recognition mechanism of a CNN is unclear, which hinders its application in many scenarios. In this paper, Self-Matching class activation mapping (CAM) is proposed to visualize what a CNN learns from SAR images to make a decision. Self-Matching CAM assigns a pixel-wise weight matrix to feature maps of different channels by matching them with the input SAR image. By using Self-Matching CAM, the detailed information of the target can be well preserved in an accurate visual explanation heatmap of a CNN for SAR image interpretation. Numerous experiments on a benchmark dataset (MSTAR) verify the validity of Self-Matching CAM.


2011 ◽  
Vol 57 (1) ◽  
pp. 37-42
Author(s):  
Krzysztof Kulpa ◽  
Mateusz Malanowski ◽  
Jacek Misiurewicz ◽  
Piotr Samczynski

Radar and Optical Images Fusion Using Stripmap SAR Data with Multilook Processing The paper presents the real-life data results of SAR and optical images data fusion. The fusion has been carried out for SAR images obtained in stripmap SAR mode using multilook processing with different methods of final image creation. The aim of the fusion was to enhance the target recognition capabilities on the Earth surface for a simple single-channel SAR receiver.


2018 ◽  
Vol 13 (2) ◽  
pp. 281-290 ◽  
Author(s):  
Wen Liu ◽  
◽  
Fumio Yamazaki

Since synthetic aperture radar (SAR) sensors onboard satellites can work under all weather and sunlight conditions, they are suitable for information gathering in emergency response after disasters occur. This study attempted to extract collapsed bridges in Iwate Prefecture, Japan, which was affected by more than 15-m high tsunamis due to the Mw 9.0 earthquake on March 11, 2011. First, the locations of the bridges were extracted using GIS data of roads and rivers. Then, we attempted to detect the collapsed or washed-away bridges using visual interpretation and thresholding methods. The threshold values on the SAR backscattering coefficients and the percentage of non-water regions were applied to the post-event high-resolution TerraSAR-X images. The results were compared with the optical images and damage investigation reports. The effective use of a single SAR intensity image in the extraction of collapsed bridges was demonstrated with a high overall accuracy of more than 90%.


2019 ◽  
Vol 11 (11) ◽  
pp. 1316 ◽  
Author(s):  
Li Wang ◽  
Xueru Bai ◽  
Feng Zhou

In recent studies, synthetic aperture radar (SAR) automatic target recognition (ATR) algorithms that are based on the convolutional neural network (CNN) have achieved high recognition rates in the moving and stationary target acquisition and recognition (MSTAR) dataset. However, in a SAR ATR task, the feature maps with little information automatically learned by CNN will disturb the classifier. We design a new enhanced squeeze and excitation (enhanced-SE) module to solve this problem, and then propose a new SAR ATR network, i.e., the enhanced squeeze and excitation network (ESENet). When compared to the available CNN structures that are designed for SAR ATR, the ESENet can extract more effective features from SAR images and obtain better generalization performance. In the MSTAR dataset containing pure targets, the proposed method achieves a recognition rate of 97.32% and it exceeds the available CNN-based SAR ATR algorithms. Additionally, it has shown robustness to large depression angle variation, configuration variants, and version variants.


2017 ◽  
Vol 3 (1) ◽  
pp. 42
Author(s):  
Roshanira Che Mohd Noor ◽  
Nur Atiqah Rochin Demong

Providing a safe and healthy workplace is one of the most effective strategies in for holding down the cost of doing construction business. It was a part of the overall management system to facilitate themanagement of the occupational health and safety risk that are associated with the business of the organization. Factors affected the awareness level inclusive of safety and health conditions, dangerous working area, long wait care and services and lack of emergency communication werethe contributed factors to the awareness level for the operational level. Total of 122 incidents happened at Telekom Malaysia Berhad as compared to year 2015 only 86 cases. Thus, the main objective of this study was to determine the relationship between safety and health factors and the awareness level among operational workers.The determination of this research was to increase the awareness level among the operational level workerswho committing to safety and health environment.


2011 ◽  
Vol 14 (2) ◽  
Author(s):  
Thomas G Koch

Current estimates of obesity costs ignore the impact of future weight loss and gain, and may either over or underestimate economic consequences of weight loss. In light of this, I construct static and dynamic measures of medical costs associated with body mass index (BMI), to be balanced against the cost of one-time interventions. This study finds that ignoring the implications of weight loss and gain over time overstates the medical-cost savings of such interventions by an order of magnitude. When the relationship between spending and age is allowed to vary, weight-loss attempts appear to be cost-effective starting and ending with middle age. Some interventions recently proven to decrease weight may also be cost-effective.


2021 ◽  
pp. 1357633X2098277
Author(s):  
Molly Jacobs ◽  
Patrick M Briley ◽  
Heather Harris Wright ◽  
Charles Ellis

Introduction Few studies have reported information related to the cost-effectiveness of traditional face-to-face treatments for aphasia. The emergence and demand for telepractice approaches to aphasia treatment has resulted in an urgent need to understand the costs and cost-benefits of this approach. Methods Eighteen stroke survivors with aphasia completed community-based aphasia telerehabilitation treatment, utilizing the Language-Oriented Treatment (LOT) delivered via Webex videoconferencing program. Marginal benefits to treatment were calculated as the change in Western Aphasia Battery-Revised (WAB-R) score pre- and post-treatment and marginal cost of treatment was calculated as the relationship between change in WAB-R aphasia quotient (AQ) and the average cost per treatment. Controlling for demographic variables, Bayesian estimation evaluated the primary contributors to WAB-R change and assessed cost-effectiveness of treatment by aphasia type. Results Thirteen out of 18 participants experienced significant improvement in WAB-R AQ following telerehabilitation delivered therapy. Compared to anomic aphasia (reference group), those with conduction aphasia had relatively similar levels of improvement whereas those with Broca’s aphasia had smaller improvement. Those with global aphasia had the largest improvement. Each one-point of improvement cost between US$89 and US$864 for those who improved (mean = US$200) depending on aphasia type/severity. Discussion Individuals with severe aphasia may have the greatest gains per unit cost from treatment. Both improvement magnitude and the cost per unit of improvement were driven by aphasia type, severity and race. Economies of scale to aphasia treatment–cost may be minimized by treating a variety of types of aphasia at various levels of severity.


Sign in / Sign up

Export Citation Format

Share Document