ImagePairs: Realistic Super Resolution Dataset via Beam Splitter Camera Rig

Joze, Hamid Reza Vaezi; Zharkov, Ilya; Powell, Karlton D.; Ringler, Carl; Liang, Luming; Roulston, Andy; Lutz, Moshe; Pradeep, Vivek

doi:10.1109/cvprw50498.2020.00267

Cited by 23 publications

(23 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The TextZoom dataset was created by Wang et al [18] using RealSR [14], and SR-RAW [16] is the first real scene text SR dataset. The text images in this collection were cut from RealSR [14] and SR-RAW [16] photos, which included stores, automobiles, gardens, and building interiors.…”

Section: 6) Textzoom [18]mentioning

confidence: 99%

“…The text images in this collection were cut from RealSR [14] and SR-RAW [16] photos, which included stores, automobiles, gardens, and building interiors. TextZoom [18] is divided into three difficulty levels: easy, medium, and severe. TextZoom [18] may be used to examine text image SR as well as text recognition.…”

Section: 6) Textzoom [18]mentioning

confidence: 99%

“…TextZoom [18] is divided into three difficulty levels: easy, medium, and severe. TextZoom [18] may be used to examine text image SR as well as text recognition.…”

Section: 6) Textzoom [18]mentioning

confidence: 99%

“…: ImagePairs [18] was suggested by Joze et al [18], which comprises 11421 LR-HR image pairs (LRHRIP) of various scenes acquired by a 5 mega pixel camera (LR) and a 20.1 mega pixel HR camera.…”

Section: 8) Imagepairs [18]mentioning

confidence: 99%

“…To capture same scene images simultaneously with two different cameras a beam splitter cube is used. But due to differences in focal length Joze et al [26] proposed pixel based aligned LRHRIP with following 4 steps i.…”

Section: 8) Imagepairs [18]mentioning

confidence: 99%

See 4 more Smart Citations

A Review on Self Learning based Methods for Real World Single Image Super Resolution

Gaikwad¹

2021

New Frontiers in Communication and Intelligent Systems

View full text Add to dashboard Cite

One of the active research issues in image processing is super resolution, which is used to boost picture resolution. The super resolution of a single image is obtained by rebuilding high-resolution (HR) pictures from low-resolution (LR) damaged pho tos (RSISR). This research examines publicly accessible datasets, RSISR assessment measures, and self-learning RSISR techniques. Comparisons are made in terms of reconstruction quality and computing efficiency utilising the self-learning RSISR technique and datasets.

show abstract

Section: 6) Textzoom [18]mentioning

confidence: 99%

Section: 6) Textzoom [18]mentioning

confidence: 99%

“…TextZoom [18] is divided into three difficulty levels: easy, medium, and severe. TextZoom [18] may be used to examine text image SR as well as text recognition.…”

Section: 6) Textzoom [18]mentioning

confidence: 99%

Section: 8) Imagepairs [18]mentioning

confidence: 99%

Section: 8) Imagepairs [18]mentioning

confidence: 99%

See 3 more Smart Citations