Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 54010 (08)
- University: Sharif University of Technology
- Department: Mechanical Engineering
- Advisor(s): Sayyadi, Hassan
- Abstract:
- Independent robots are equipped with various sound, inertia and visual sensors for decision making. Vision is an attractive sensor due to its non-invasive nature, passivity, and high information content. In natural environments, visual noises such as snow, rain, and dust distort images. in underwater environments, factors such as refraction and absorption of light suspended particles in the water, and color distortion affects the quality of visual data, resulting in noisy and distorted images. As a result, the autonomous underwater vehicles that rely on vision (AUVs) are challenged, resulting in poor performance. To improve the input to the visual algorithm for tracking the pipeline, in this research we proposed a method to improve the quality of underwater images using GAN generators. For this purpose, by designing four different models, we selected the most optimal model in terms of runtime with the amount of improved image quality. the output of the Improved image from the selected GAN network is the input of our proposed tracking algorithm and increase the safety and reliability of the optical sensors. In order to track underwater lines, we used a two-step algorithm, which made the program lighter and therefore more real-time. Tracking results for three different environments in two modes with improved quality and without The quality improvement was compared. The results show that the performance of the tracking algorithm using quality image input is significantly improved and is appropriate to the real situation
- Keywords:
- Computer Vision ; Autonomous Underwater Vehicle ; Tracking ; Image Enhancement ; Generative Adversarial Networks
Digital Object List
-
محتواي کتاب
- view
Bookmark