Automatic tracing and extraction of text‐line and word segments directly in JPEG compressed document images

Rajesh, Bulla; Javed, Mohammed; Nagabhushan, P.

doi:10.1049/iet-ipr.2019.1437

Cited by 5 publications

(7 citation statements)

References 35 publications

(58 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, If f’(k, l) has further reduced with y% then the total reduction of size is (x + y + (x+y)/100) . Two such approaches [ 8 , 29 ] are considered from the literature to apply them for the segmentation of JPEG compressed documents into text-lines and words. Moreover, though these types of approaches in the literature are proposed for segmenting large objects, the proposed models in this paper are applied to achieve more faster and smoother segmentation of text-line and word boundaries.…”

Section: Proposed Modelmentioning

confidence: 99%

“…Based on this fact, from a typical DCT image of size f (i, j) all the DC values from each 8 × 8 DCT block are extracted and mapped onto the another file by preserving their positions and order as it is. This newly mapped file is called the DC_Reduced image [ 29 ]. Because of this mapping the original image of size f (i, j) gets reduced to a low resolution of size f (i/8, j/8).…”

Section: Proposed Modelmentioning

confidence: 99%

“…Because of locating the expected separators on the low resolution image, this method is more faster when compared to the DCT image. The DC_Reduced model in [ 29 ] is proposed for segmenting the text-lines and words in handwritten documents. But, since printed text-lines and words are very close in nature, and one full DCT block is replaced by just one DC value the DC_Reduced image of printed text document does not show the separation gaps clearly for estimating the clear boundaries of text-lines and words.…”

Section: Proposed Modelmentioning

confidence: 99%

“…Similarly the second approach is Fast and Smooth Segmentation (FastSS-2) based on analyzing AC coefficients in selected 8 × 8 DCT blocks and without any decompression. Further, two Faster algorithms are proposed for More Faster and Smooth Segmentation (MFastSS-1 and MFastSS-2) in the JPEG compressed domain using the idea of extracting low resolution images from compressed data by [ 8 , 29 ]. Low resolution images are generated by mapping the DC signal (called as DC Reduced image) and encoded DCT coefficients (called as ECM image).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

FastSS: Fast and smooth segmentation of JPEG compressed printed text documents using DC and AC signal analysis

Rajesh

Javed

Nagabhushan

2022

Multimed Tools Appl

Self Cite

View full text Add to dashboard Cite

With the surge of COVID-19 pandemic, the world is moving towards digitization and automation more than it was presumed. The Internet is becoming one of the popular mediums for communication, and multimedia (image, audio, and video) combined with data compression techniques play a pivotal role in handling a huge volume of data that is being generated on a daily basis. Developing novel algorithms for automatic analysis of compressed data without decompression is the need of the present hour. JPEG is a popular compression algorithm supported in the digital electronics world that achieves compression by dividing the whole image into non-overlapping blocks of 8 × 8 pixels, and subsequently transforming each block using Discrete Cosine Transform (DCT). This research paper proposes to carry out Fast and Smooth Segmentation (FastSS) directly in JPEG compressed printed text document images at text-line and word-level using DC and AC signals. From each 8 × 8 block, DC and AC signals are analyzed for accomplishing Fast and Smooth segmentation, and subsequently, two Faster segmentation (MFastSS) algorithms are also devised using low resolution-images generated by mapping the DC signal (DC Reduced Image) and encoded DCT (ECM Image) coefficients separately. Proposed models are tested on various JPEG compressed printed text document images created with varied space and fonts. The experimental results have demonstrated that the direct analysis of compressed streams is computationally efficient, and has achieved speed gain more than 90% when compared to uncompressed domains.

show abstract

Section: Proposed Modelmentioning

confidence: 99%

Section: Proposed Modelmentioning

confidence: 99%

Section: Proposed Modelmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

FastSS: Fast and smooth segmentation of JPEG compressed printed text documents using DC and AC signal analysis

Rajesh

Javed

Nagabhushan

2022

Multimed Tools Appl

Self Cite

View full text Add to dashboard Cite

show abstract

“…Here, it should be noted that the techniques that extract words from document images might be erroneous sometimes and can add some computational overhead if a segmentation-based approach is adopted. However, the state-of-the-art word extraction methods [ 18 , 19 , 20 ] that perform the tasks with good efficiency on complex documents, while consuming less time, could be used to get rid of these issues. For a similar reason, text line-based techniques are computationally more expensive when compared to word-based methods.…”

Section: Introductionmentioning

confidence: 99%

Hough Transform-Based Angular Features for Learning-Free Handwritten Keyword Spotting

Kundu

Malakar²,

Geem

et al. 2021

Sensors

View full text Add to dashboard Cite

Handwritten keyword spotting (KWS) is of great interest to the document image research community. In this work, we propose a learning-free keyword spotting method following query by example (QBE) setting for handwritten documents. It consists of four key processes: pre-processing, vertical zone division, feature extraction, and feature matching. The pre-processing step deals with the noise found in the word images, and the skewness of the handwritings caused by the varied writing styles of the individuals. Next, the vertical zone division splits the word image into several zones. The number of vertical zones is guided by the number of letters in the query word image. To obtain this information (i.e., number of letters in a query word image) during experimentation, we use the text encoding of the query word image. The user provides the information to the system. The feature extraction process involves the use of the Hough transform. The last step is feature matching, which first compares the features extracted from the word images and then generates a similarity score. The performance of this algorithm has been tested on three publicly available datasets: IAM, QUWI, and ICDAR KWS 2015. It is noticed that the proposed method outperforms state-of-the-art learning-free KWS methods considered here for comparison while evaluated on the present datasets. We also evaluate the performance of the present KWS model using state-of-the-art deep features and it is found that the features used in the present work perform better than the deep features extracted using InceptionV3, VGG19, and DenseNet121 models.

show abstract

Deep Learning-Based Image Retrieval in the JPEG Compressed Domain

Temburwar

Rajesh

Javed

2022

Lecture Notes in Electrical Engineering

View full text Add to dashboard Cite

Automatic tracing and extraction of text‐line and word segments directly in JPEG compressed document images

Cited by 5 publications

References 35 publications

FastSS: Fast and smooth segmentation of JPEG compressed printed text documents using DC and AC signal analysis

FastSS: Fast and smooth segmentation of JPEG compressed printed text documents using DC and AC signal analysis

Hough Transform-Based Angular Features for Learning-Free Handwritten Keyword Spotting

Deep Learning-Based Image Retrieval in the JPEG Compressed Domain

Contact Info

Product

Resources

About