|
Document Type |
: |
Thesis |
|
Document Title |
: |
A HARDWARE & SOFTWARE CO-DESIGN FOR VISUAL PERCEPTION USING SEMI- GLOBAL BLOCK MATCHING-BASED DEPTH ESTIMATION FROM REAL-TIME STEREO IMAGES تصميم مشترك للأجهزة والبرمجيات للإدراك البصري باستخدام تقدير العمق المستند الى مطابقة الكتلة شبه العالمية من صور الاستريو في الوقت الفعلي |
|
Subject |
: |
Faculty of Engineering |
|
Document Language |
: |
Arabic |
|
Abstract |
: |
In robot navigation, automated driverless car operation and other futuristic technologies requiring automated environmental visualization, visual awareness is a critical role. Depth estimates are a crucial and offensive aspect of visual experience researchers and creators since all are very challenging Precision and processing objectives in real time. There are already two modalities of parallel hardware this algorithm, i.e. LiDar and stereoscopic cameras, is known to implement. For their respective autonomous vehicle ventures, Google and Tesla adopted two. However, due to its exorbitant price, the former platform is discouraged. Finally yet importantly, with cheaper stereo cameras, enormous computer resources still needed to create an Environmental depth chart. In this thesis, we suggest a hardware device code; sign for the usage of commercially accessible stereo cameras for visual perception applications the breadth assessment used to fulfil the FPGA's concurrent processing capacity the stereoscopy algorithm needs a heavy computing load in real time. We, in particular, the algorithm for depth estimation will implemented with the semi-global block matching recognized for its FPGA implementation appropriateness. The sensory recognition device suggested integrates Open CV applications library Open Source machine vision, with built hardware accelerator for a strong and scalable depth assessment, Code design framework hardware-software. The results of proposed algorithm shows high improvement then previous schemes in literature. |
|
Supervisor |
: |
Dr. Mohamed Farhan |
|
Thesis Type |
: |
Master Thesis |
|
Publishing Year |
: |
1443 AH
2022 AD |
|
Added Date |
: |
Monday, January 23, 2023 |
|
Researchers
| عمر فرحان القرني | Al-Qarni, Omar Farhan | Researcher | Master | |
|