This paper provides a comprehensive review of SURF (speeded up robust features) feature descriptor, commonly used technique for image feature extraction. The SURF algorithm has obtained significant popularity because to its robustness, efficiency, and invariance to various image transformations. In this paper, an in-depth analysis of the underlying principles of SURF, its key components, and its use in computer vision tasks such as object recognition, image matching, and 3D reconstruction are proposed. Furthermore, we discuss recent advancements and variations of the SURF algorithm and compare it with other popular feature descriptors. Through this review, the aim is to provide a clear understanding of the SURF feature descriptor and its significance in the area of computer vision.

Objective: to achieve accurately and rapidly the mapping of agricultural land use and crop distribution at the township scale. Methods: this study, based on specific methods, such as, time-series remote sensing index threshold classification and maximum likelihood, classifies each land use type and extracts crop spatial information, under the guidance of Sentinel-2A remote sensing images, to carry out agricultural land use mapping at township scale. And the mapping concerned will be verified by comparing with an agricultural spatial information map of a 0.5 m resolution, which is based on WorldVieW-2 fused images. Results: (1) the area accuracy of grain and oil crop land, vegetable land, agricultural facilities land and garden land is fairly good, with 92.93%, 98.98%, 95.71% and 95.14% respectively, and within 8% variation from actual area; (2) the spatial information of plot boundary, farmland road network, and canal network produced by OSM road data and historical high-resolution images was overlayed with the classification results of Sentinel-2A multi-spectral image for mapping, which can improve the accuracy of plot boundary information of classification results for the image with 10 m resolution. Conclusions: the use of multi-source information fusion method, agricultural land use and crop distribution space big data produced by Sentinel-2A optical image, can effectively improve the accuracy and timeliness of land use mapping at the township scale, to provide technical reference for the application of remote sensing big data to carry out agricultural landscape analysis at the township scale, optimization and adjustment of agricultural structure, etc.

Modelling and simulation have now become standard methods that serve to cut the economic costs of R&D for novel advanced systems. This paper introduces the study of modelling and simulation of the infrared thermography process to detect defects in the hydroelectric penstock. A 3-D penstock model was built in ANSYS version 19.2.0. Flat bottom holes of different sizes and depths were created on the inner surface of the model as an optimal scenario to represent the subsurface defect in the penstock. The FEM was applied to mimic the heat transfer in the proposed model. The model’s outer surface was excited at multiple excitation frequencies by a sinusoidal heat flux, and the thermal response of the model was presented in the form of thermal images to show the temperature contrast due to the presence of defects. The harmonic approximation method was applied to calculate the phase angle, and its relationship with respect to defect depth and defect size was also studied. The results confirmed that the FEM model has led to a better understanding of lock-in infrared thermography and can be used to detect subsurface defects in the hydroelectric penstock.

In agriculture, crop yield and quality are critical for global food supply and human survival. Challenges such as plant leaf diseases necessitate a fast, automatic, economical, and accurate method. This paper utilizes deep learning, transfer learning, and specific feature learning modules (CBAM, Inception-ResNet) for their outstanding performance in image processing and classification. The ResNet model, pretrained on ImageNet, serves as the cornerstone, with introduced feature learning modules in our IRCResNet model. Experimental results show our model achieves an average prediction accuracy of 96.8574% on public datasets, thoroughly validating our approach and significantly enhancing plant leaf disease identification.