University of Twente Student Theses

Login
This website will be unavailable due to maintenance December 1st between 8:00 and 12:00 CET.

Blind Image Quality Assessment of Smartphone-captured Images in the Wild

Chandra Mohan, Tejas (2021) Blind Image Quality Assessment of Smartphone-captured Images in the Wild.

[img] PDF
10MB
Abstract:Real-world images captured using an imaging device suffers from distortion while capturing, processing, or storage. These distortions in images affect their visual quality, rendering them unusable for further processing. This thesis concentrates on images captured by a smartphone from behind a car's windshield. The objective is to classify these images into good quality and bad quality employing deep learning models focusing on Image Quality Assessment. This paper provides an overview of recent developments in Blind Image Quality Assessment (BIQA) using deep learning and the available standard datasets. Specifically, three recent BIQA models are selected to evaluate these images and quantify them as good and bad based on their image quality. Further research is conducted on an ensemble of these BIQA models for the same task. Later, a classification approach is explored consisting of three transfer learning models to classify the images as good quality and bad quality. An ensemble comprising of these models is built. The test results show that the ensemble combination comprising of two BIQA models delivers the highest accuracy towards rightly classifying images as good quality and bad quality.
Item Type:Essay (Master)
Clients:
CamenAI, Utrecht, The Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Embedded Systems MSc (60331)
Link to this item:https://purl.utwente.nl/essays/89267
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page