KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Search Results

Now showing 1 - 6 of 6

Some of the metrics are blocked by your
consent settings
Enhancing deep learning–based railway inspection via PSO-guided brightness–contrast optimization
(2026-03-01)
Songthai, Maethinee
;
Kaitwanidvilai, Somyot
Reliable operation of electrified railway systems depends critically on the pantograph–catenary system (PCS). Existing inspection practices are largely manual or periodic and remain vulnerable to low illumination, background clutter, and thin structural components, limiting robustness and scalability. Although vision-based deep learning is promising, performance often degrades in low-light and complex scenes, while conventional enhancement (e.g., CLAHE) provides limited and inconsistent improvements. This study proposes an illumination-aware PCS inspection framework that integrates Particle Swarm Optimization (PSO)–guided brightness–contrast optimization with a lightweight YOLO detector. A version benchmark selected YOLOv9t as the backbone, achieving the best overall performance (Precision = 0.947, F1-score = 0.913) compared with YOLOv8n and YOLOv11n. Although YOLOv11n has lower FLOPs, PCS inspection is accuracy-critical because detection errors propagate to event-frequency counting and lateral-deviation assessment; therefore, YOLOv9t was fixed for subsequent experiments. PSO is employed to estimates a single dataset-level global enhancement parameter set, improving robustness while maintaining computational efficiency for early-stage field deployment under limited annotated data and edge hardware constraints. The framework was evaluated on a small yet diverse visible-light dataset spanning day/evening/night conditions and challenging locations (station roofs, bracket regions, and transition sections). Comparative evaluation across three configurations—YOLOv9 baseline, YOLOv9 with CLAHE, and YOLOv9 with PSO-tuned brightness/contrast—achieved detection rates of 90%, 79%, and 99%, respectively, with the largest gains on thin, low-contrast structures (contact and messenger wires). The optimized pipeline further enables reliable estimation of lateral wire deviation in compliance with EN 50367 (≤200 mm), supporting safety-critical inspection in low-light and complex-background conditions.
Some of the metrics are blocked by your
consent settings
Adaptive Force Control Using a Standard Deviation-based Hybrid Approach
(2023-01-01)
Songthai, Maethinee
;
Yanyong, Sarucha
;
Kaitwanidvilai, Somyot
A good tracking response is essential for a force control system to achieve high performance in object handling. One of the significant challenges of the force control system is that the plant dynamics are directly dependent on the environment and the mechanical properties of the object. It's well-known that the mechanical properties of an object in manufacturing vary based on the type of product, gripper tools, and environment; hence, a non-adaptive controller may not be efficiently adopted. To address this issue, this paper proposes impedance control with hybrid adaptive algorithms to ensure the actuator system tracks the desired force command accurately. Particle Swarm Optimization (PSO) is adopted to adapt the impedance control parameters, thereby applying the capabilities of the learning system. A hybrid adaptive force control with a fitness function of the standard deviation and summation of error is proposed. The tracking response of the conventional adaptive controller was examined and compared with the proposed controller. The simulation results demonstrate the effectiveness of the proposed system.
Some of the metrics are blocked by your
consent settings
Estimation of the organic film thickness on stainless steel using image intensity
(2018-08-14)
Songthai, Maethinee
;
Suesut, Taweepol
;
Nunak, Navaphattra
This research proposes an estimation of thickness of organic film coating on a Stainless Steel surface (SS) and a clear microscope slide using a visible optical method with the image intensity level analysis. Sweetened Condensed Milk (SCM) was used as a representative of the organic film. The uncertainty of inverse calibration equations was evaluated to determine the reliability of this technique. The results showed that the image intensity increased with the increasing of the film thickness in a linear relationship with a high coefficient of determination (R<sup>2</sup> > 0.97). Uncertainty of both inverse calibration equations was low and acceptable. It could be concluded that this technique could be used to estimate the thickness of the organic translucent film coating directly on a SS or a clear slide with the limitation of approximately 1.0 mm thickness using the reported linear inverse calibration equations.
Some of the metrics are blocked by your
consent settings
Investigation of soy milk deposited on stainless steel by infrared thermography
(2018-01-01)
Suesut, Taweepol
;
Songthai, Maethinee
;
Nunak, Navaphattra
This paper proposes a real-time investigation of soy milk deposited on stainless steel surface (SS) during the heating process by considering the relationship between emissivity and mass of soils on the surface using a thermal image processing technique. The understanding on organic fouling behavior is an important step leading to the optimum cleaning operations. The mass of soy milk deposited on SS during heating process at the temperature of 75°C for 180 min was measured by the weighing method and compared to the emissivity values analyzed from the infrared thermography in real-time. Two different types of stainless steel grades (AISI 304 and 316) with various average surface roughness (R<inf>a</inf>) values (0.4, 0.8 and 3.2 µm) were carried out. Emissivity values of sample plates which soil deposited on the surface, were obtained from the real-time processing of a thermo-map of soil film compared to the temperature of a reference surface (known emissivity value). Applying this technique to all conditions, it was found that the increasing of emissivity values of sample plates as the amount of soil film on SS increased could be detected in real-time for both SS grades. The emissivity of SS having soil on the surface was higher than a clean SS since the soil film on the SS caused the roughness of surface changed. It could also detect that emissivity values of both SS grades had no significantly difference at the same R<inf>a</inf>. From the detection of soy milk deposited on SS using the real-time thermal image processing acquired from infrared camera during heating process, it could be concluded that the proposed technique was possible to investigate the accumulation of soy milk and other soil types on the surface.
Some of the metrics are blocked by your
consent settings
Estimation of thermal conductivity and thermal diffusivity of biological materials using thermographic camera
(2014-01-01)
Nunak, Navaphattra
;
Sritham, Eakasit
;
Songthai, Maethinee
;
Suesut, Taweepol
This study proposes a method to estimate thermal conductivity and thermal diffusivity using a thermoelectric device and a thermographic camera. The experimental results were compared with others literatures. Three different materials including acrylic, apple, and potato were tested. The proposed method has proved to be effective. The experimental results for all selected materials are in comparable range with literature values. The proposed method can be a useful tool for determining thermal properties of solid biological materials with low thermal conductivity and diffusivity values.
Some of the metrics are blocked by your
consent settings
Volume estimation of symmetrical object using laser light sectioning
(2014-01-01)
Gulphanich, Suphan
;
Songthai, Maethinee
;
Suesut, Taweepol
This paper presents a method to estimate volume from three-dimensional characteristics of symmetrical objects by computer vision system using laser light sectioning. The camera captures the image of laser line projected on the object moving on a conveyor. The captured images can be estimated the widths and heights at various locations along the object. The area of cross section can be integrated to be volume of the object. The image data is extracted from the structured laser light projecting on the surface. After image pre-processing, the laser profile is calculated the center of gravity in order to extract the image coordinate. The homography transformation is used to transform the extracted coordinate to the real world distance (in metric system). A metric vision based on light sectioning for estimating the volume of symmetrical shapes (rectangular shape and circular shape) of the object were tested. The accuracy of estimation can apply in the automatic size grading on the conveyer as well.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results