Accuracy of OCR is often marred by the poor quality of the input document images. Generally this performance degradation is attributed to the resolution and quality of scanning. This calls for special efforts to improve the quality of document images before passing it to the OCR engine. One compelling option is to super-resolve these low resolution document images before passing them to the OCR engine. In this work we address this problem by super-resolving document images using Generative Adversarial Network (GAN). We propose a super resolution based preprocessing step that can enhance the accuracies of the OCRs (including the commercial ones). Our method is specially suited for printed document images. We validate the utility in wide variety of document images (where fonts, styles, and languages vary) without any pre-processing step to adapt across situations. Our experiments show an improvement upto 21% in accuracy OCR on test images scanned at low resolution. One immediate application of this can be in enhancing the recognition of historic documents which have been scanned at low resolutions.

Enhancing OCR Accuracy with Super Resolution

This paper addresses the problems encountered during digitization and preservation of inscriptions such as perspective distortion and minimal distinction between foreground and background. In general inscriptions possess neither standard size and shape nor colour difference between the foreground and background. Hence the existing methods like variance based extraction and Fast ICA based analysis fail to extract text from these inscription images. Natural gradient flexible ICA (NGFICA) is a suitable method for separating signals from a mixture of highly correlated signals, as it minimizes the dependency among the signals by considering the
slope of the signal at each point. We propose an NGFICA based enhancement of inscription images. The proposed method improves word and character recognition accuracies of the OCR system by 65.3% (from 10.1% to 75.4%) and 54.3% (from 32.4% to 86.7%), respectively.

/pdf/ngfica-based-digitization-of-historic-inscription-images-42s52ls99v.pdf

NGFICA Based Digitization of Historic Inscription Images

Over the past decade the document analysis and processing related to camera based document images has gained the interest of research community. Nowadays, cameras are easily available in the smart phones that can be carried in the small space of our pockets while being lightweight, portable and relieving us from the burden of walking down to a scanner for a digital copy of a document. But even though capturing a document image through a phone camera appears simple, the chances of obtaining a perfect picture are scanty. As when the picture is captured in an unconstrained environment, there are chances of degradation to creep in that will hamper the visual quality of the document image which further effect the readability(in terms of OCR accuracy). Low quality documents give poor results. Document images contain various degradations such as blur, uneven illumination, perspective distortion, low resolution, smear etc. Quality enhancement is helpful to recognize a camera captured document more accurately and if not completely removing the degradations, it can be used for suppressing them and making the text more readable. This paper evaluates the performance of various deblurring techniques for noisy and blurred camera captured documents.

Image processing based degraded camera captured document enhancement for improved OCR accuracy

This paper addresses the problems encountered during digitization and preservation of inscriptions such as perspective distortion and minimal distinction between foreground and background. In general inscriptions neither possess standard size and shape nor colour difference between the foreground and background. Hence the existing methods like variance based extraction and Fast-ICA based analysis fail to extract text from these inscription images. Natural gradient Flexible ICA (NGFICA) is a suitable method for separating signals from a mixture of highly correlated signals, as it minimizes the dependency among the signals by considering the slope of the signal at each point. We propose an NGFICA based enhancement of inscription images. The proposed method improves word and character recognition accuracies of the OCR system by 65.3% (from 10.1% to 75.4%) and 54.3% (from 32.4% to 86.7%) respectively.

Enhancement of inscription images

With the burst of low-cost and versatile mobile phones, people have more choices to digitize papers simply by capturing the text images of interest and put them through Optical Character Recognition (OCR) systems. However, the task is much more challenging when it comes to the issue of recognition of camera-based text images taken at remote distances. Those images, which often come up in small font and low resolution texts, results in poor readability in OCR. This paper presents a preliminary study on the human eye's perception of English document images, taken by mobile phone at various distances. In addition, a text enhancement technique using the combination of power-law transformation and morphological bottom hat filtering, following by morphological reconstruction and interpolation is also addressed. The results achieve an improved performance in OCR recognition rate.

A study about the reconstruction of remote, low resolution mobile captured text images for OCR

An Independent Component Analysis (ICA)-based image enhancement technique is presented to improve the accuracy for machine reading of camera-based images. Images of inscriptions that are normally engraved on stones or other durable materials and found at the sites of historical monuments are taken as a reference for conducting the present experiments. Significant improvement in recognition rate of a commercial OCR system shows the potential of the proposed ICA-based method. It improves word and character recognition accuracies of the OCR system by 68.6% (from 11.2% to 79.8%) and 57.3% (from 34.8% to 92.1%), respectively.

A. Maity

Papers

Machine reading of camera-held low quality text images: An ICA-based image enhancement approach for improving OCR accuracy