Image refinement using skew angle detection and correction for Indian license plates

Arulmozhi, K; Perumal, S. Arumuga; Priyadarsini, C S Thanooja; Nallaperumal, Krishnan

doi:10.1109/iccic.2012.6510316

“…for skew angle detection and free skew correction, [3-8, 15, 39]; but in this study, we adopted Hough Transform proposed by Paul Hough in 1962 because of its comparative robustness and high accuracy even though it increases computational time. Besides, it has been used in vehicle license plate recognition system to remove skew angle [32]. The tradeoff therefore is between achieving a low accuracy with fast processing or a high accuracy with longer computation time.…”

Section: Several Researchers Have Proposed Diverse Methodsmentioning

confidence: 99%

Image Purification Technique for Myanmar OCR Applying Skew Angle Detection and Free Skew

Lwin

¹

,

Wu

²

2019

IJSRST

View full text Add to dashboard Cite

Optical Character Recognition (OCR) is a technology widely adopted for automatic translation of hardcopy text to editable text. The language dependence of the technology makes it far less developed for less popular languages like Myanmar language. Also, the uniqueness and complexity of the Myanmar text system such as touching and complex characters have continued to pose serious challenges to several OCR investigators. In this paper, we propose a new technique to development Myanmar OCR system. Our technique implement skew angle detection and free skew, noisy border correction, extra page elimination, line segmentation from scanned images of Myanmar text. Performance of the proposed method is tested with 430 documents comprising different printed and handwritten Myanmar text of various fonts, sizes, multi-column, tables, stamps or photos, background effects. Our method give an accuracy of 100% for line segmentation and 99.92% for skew angle detection and free skew. The ability of our method to effectively implement global and local skew angle detection, free skew and line segmentation in different handwritten and digital text images of the Myanmar character set with high accuracies confirms the robustness of the technique, its reliability and its suitability for application in many other related languages.

show abstract

“…These methods are designed to detect the tilt which are less than 35 0 . These approaches are based on Principle Component Analysis [2], Fourier Transform [3], Hough Transform [4], Nearest neighbor connectors [5], Extreme points [6] and Moments [7,8]. But most approaches fail to detect the orientations of 90 0 , 180 0 and 270 0 .…”

Section: Literature Surveymentioning

confidence: 99%

Detection of Oriented Number Plate in Vehicle Using Autocorrection Feature From Gray Level Co-Occurence Matrix

M.N¹,

S.J²,

Vasudev³

2015

Computer Science &Amp; Information Technology ( CS &Amp; IT )

0

View full text Add to dashboard Cite

The efficiency of an automatic number plate recognition system depends directly on the proper effective preprocessing of the number plate. The OCRs available for recognition are capable of reading the number plates which are in proper orientation of 00. In many situations the vehicle number plates captured may be in any different orientation like 900, 1800 and 2700. These orientations in number plates are due to declamping of number plate at one end or toppling of vehicle. Such differently oriented number plates cannot be subjected for reading by OCRs and such situations require the system to detect the direction of orientation and correct the same before subjecting the same for reading. This paper proposes a work to detect the orientation of segmented number plates from vehicle image using autocorrelation feature from gray level cooccurrence matrix. A good volume of training samples are generated synthetically to train the system and the system is tested using sufficient test samples. The results of system shows an overall efficiency of 65.61% and performs an essential preprocessing in an automatic number plate recognition system.

show abstract

“…These methods are designed to detect the tilt which are less than 35 0 . These approaches are based on Principle Component Analysis [10], Fourier Transform [11], Hough Transform [12], Nearest neighbor connectors [13], Extreme points [14] and Moments [15,16]. But most approaches fail to detect the orientations of 90 0 , 180 0 and 270 0 .…”

Section: Literature Surveymentioning

confidence: 99%

“…The skew causes major effect on character segmentation and recognition. Good number of methods have been proposed by many researchers for the detection of skew in vehicle number plate [1] in order to increase the performance of OCR system such as using principal component analysis [2], Fourier transform [3], Hough transform [4], nearest neighbour connectivity [5], method of extreme points [6] and moments [7,8].…”

Section: Introductionmentioning

confidence: 99%

Detection of Oriented and Skewed Number Plate in Vehicle Using Hybrid Approach

M.N¹,

S.J²,

Vasudev³

2015

IJMA

0

View full text Add to dashboard Cite

The efficiency of an automatic number plate recognition system depends directly on the proper effective preprocessing of the number plate. The OCRs available for recognition are capable of reading the number plates which are in proper orientation of 0 0. In many situations the vehicle number plates captured may be in any different orientation like 90 0 , 180 0 and 270 0. These orientations in number plates are due to declamping of number plate at one end or toppling of vehicle. Such differently oriented number plates cannot be subjected for reading by OCRs and such situations require the system to detect the direction of orientation and correct the same before subjecting the same for reading. The efficiency of an automatic vehicle number plate recognition is high if the necessary preprocessing methods give results effectively. Also an essential and important preprocessing in automatic number plate recognition system is correction of skewed number plates in vehicle images which is mainly due to position of camera while capturing the vehicle image. The skewed number plate affects badly on the accurate character segmentation and recognition. Once the number plate is segmented from the vehicle image, the plate has to be checked for skewness and the same has to be corrected for future processing. This paper proposes a work to detect the orientation and skewed of segmented number plates from vehicle image using hybrid approach combination of Autocorrelation and Radon transform. A good volume of training samples are generated synthetically to train the system and the system is tested using sufficient test samples. The results of system shows an overall efficiency 65.05% of oriented detection, efficiency of 85.71% skewed number plate and performs an essential preprocessing in an automatic number plate recognition system.

show abstract

Image refinement using skew angle detection and correction for Indian license plates

Cited by 13 publications

References 2 publications

Image Purification Technique for Myanmar OCR Applying Skew Angle Detection and Free Skew

Image Purification Technique for Myanmar OCR Applying Skew Angle Detection and Free Skew

Detection of Oriented Number Plate in Vehicle Using Autocorrection Feature From Gray Level Co-Occurence Matrix

Detection of Oriented and Skewed Number Plate in Vehicle Using Hybrid Approach

Contact Info

Product

Resources

About