Log-polar Transformation as a Tool for Text Skew Estimation

Brodić, Darko; Milivojević, Zoran N.; Medvedeva, Aleksandra

doi:10.5755/j01.eee.19.2.3471

Cited by 8 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Log-polar transformations have been used to estimate skewness of text in an image [4], to detect road signs in an image [5]. Amorim et al [6] demonstrated that log-polar preprocessing can assist with rotation invariance in CIFAR10 images, but did not demonstrate scale invariance, data processing efficiency, nor were they able to demonstrate rotation invariance with the character-based datasets (e.g.…”

Section: Related Workmentioning

confidence: 99%

Human eye inspired log-polar pre-processing for neural networks

Remmelzwaal¹,

Mishra²,

Ellis³

2019

Preprint

View full text Add to dashboard Cite

In this paper we draw inspiration from the human visual system, and present a bio-inspired pre-processing stage for neural networks. We implement this by applying a log-polar transformation as a pre-processing step, and to demonstrate, we have used a naive convolutional neural network (CNN). We demonstrate that a bio-inspired pre-processing stage can achieve rotation and scale robustness in CNNs. A key point in this paper is that the CNN does not need to be trained to identify rotation or scaling permutations; rather it is the log-polar pre-processing step that converts the image into a format that allows the CNN to handle rotation and scaling permutations. In addition we demonstrate how adding a log-polar transformation as a pre-processing step can reduce the image size to 20% of the Euclidean image size, without significantly compromising classification accuracy of the CNN. The pre-processing stage presented in this paper is modelled after the retina and therefore is only tested against an image dataset. 1

show abstract

Section: Related Workmentioning

confidence: 99%

Human eye inspired log-polar pre-processing for neural networks

Remmelzwaal¹,

Mishra²,

Ellis³

2019

Preprint

View full text Add to dashboard Cite

show abstract

“…The Cartesian space coordinates x and y are converted into polar coordinates radius r and angleθ . Their mapping is as Eq.5 [5,6,7]. Furthermore, the log transformation is obtained by Eq.6.…”

Section: Well Log Image Features Extractionmentioning

confidence: 99%

An Effective Multi-Feature Extraction Method for Manually Plotted Well-Log Curve Line

Xia

Gao

Zhang

2014

AMM

View full text Add to dashboard Cite

An effective multi-feature extraction method for well-log curve line plotted manually. Several log curve line features are used to extract the well log curve information in the proposed method: gray value similarity feature, connected region Length feature, projection feature and distance and angular feature. The experimental results show the good effect of the method for the well log curve information extraction.

show abstract

“…After finding a solution in that space, the information is translated into image space in the form of the text skew estimation. They include methods like Hough transform [2], [31], [33], Fourier transform [26], Radon transform [1], [16], log-polar transform [6].…”

Section: Introductionmentioning

confidence: 99%

Text Skew Detection Using Combined Entropy Algorithm

Brodić

Milivojević

2017

ITC

View full text Add to dashboard Cite

This paper proposes the algorithm for text skew estimation based on the combined entropy calculation. The method consists of three steps. In the first step, it calculates the horizontal and vertical projection profiles. In the second step, the horizontal and vertical entropy for the rough angles is calculated. In the last step, the horizontal and vertical entropy calculation for the smooth angles is performed. The calculated entropy creates the two cost functions: horizontal and vertical. The position where each of cost functions has an absolute minimum represents horizontal and vertical estimated text skew angle. In the last step, it estimates the text skew angle as a mean of horizontal and vertical estimated text skew angles. The functionality, correctness and robustness of the proposed algorithm is investigated by the experiment, which is based on DISEC'13 document database. The obtained results are promising. They confirm that the method correctly estimates text skew angle.

show abstract

Log-polar Transformation as a Tool for Text Skew Estimation

Cited by 8 publications

References 8 publications

Human eye inspired log-polar pre-processing for neural networks

Human eye inspired log-polar pre-processing for neural networks

An Effective Multi-Feature Extraction Method for Manually Plotted Well-Log Curve Line

Text Skew Detection Using Combined Entropy Algorithm

Contact Info

Product

Resources

About