OCR Accuracy Improvement through a PDE-Based Approach

Abstract

This paper focuses on improving the Optical Character Recognition (OCR) system's accuracy by restoring damaged character through a PDE (Partial Differential Equation)-based approach. This approach, proposed by D. Tschumperlé, is an anisotropic diffusion approach driven by local tensors fields. Actually, such approach has many useful properties that are relevant for use in character restoration. For instance, this approach is very appropriate for the processing of oriented patterns which are major characteristics of textual documents. It incorporates both edge enhancing diffusion that tends to preserve local structures during smoothing and coherence-enhancing diffusion that processes oriented structures by smoothing along the flow direction. Furthermore, this tensor diffusion-based approach compared to the existing sate of the art requires neither segmentation nor training steps. Some experiments, done on degraded document images, illustrate the performance of this PDE-based approach in improving both of the visual quality and the OCR accuracy rates for degraded document images.

Pre-Processing of Degraded Printed Documents by Non-local Means and Total Variation

“…NLmeans is more effective when small parts of characters are degraded since the method can restore degraded parts from neighboring image parts. Future work will consist of comparing the proposed enhancement methods to classical ones such as PDEbased approaches [8]. The TV β parameter must be adapted to character resolution in order to preserve character details.…”

“…In [16], both character repairation and binarization are performed through MRF modeling. Restoring character edges by PDE-based approaches has been proposed in [8]. This approach regularizes a document image using an anisotropic diffusion filtering.…”

Enhancement of historical printed document images by combining Total Variation regularization and Non-local Means filtering

“…As mentioned in Section 1, they can also be used for restoring images, including document images [14]. A popular PDE approach is the Perona-Malik anisotropic diffusion approach [30].…”

“…The water flow model of [21] can extract the different background layers of a document while binarizing it. Restoring character edges by PDE-based (Partial Differential Equations) approaches has been proposed in [14]. This approach regularizes a document image using anisotropic diffusion filtering through an iterative process.…”

RSLDI: Restoration of single-sided low-quality document images

“…In these cases, the appropriate use of an inverse diffusion model can result in very good and direct restoration of these types of physical diffusion. Moreover, the denoising nature of diffusion models provides clear output, without the need for extra denoising postprocessing [19]. The main point about diffusion methods is their great similarity with the physical processes of degradation [20].…”