Charalambos Strouthopoulos

Multithresholding of color and gray-level images through a neural network technique

One of the most frequently used methods in image processing is thresholding. This can be a highly efficient means of aiding the interpretation of images. A new technique suitable for segmenting both gray-level and color images is presented in this paper. The proposed approach is a multithresholding technique implemented by a Principal Component Analyzer (PCA) and a Kohonen Self-Organized Feature Map (SOFM) neural network. To speedup the entire multithresholding algorithm and reduce the memory requirements, a sub-sampling technique can be used. Several experimental and comparative results exhibiting the performance of the proposed technique are presented. q 2000 Elsevier Science B.V. All rights reserved.

Text extraction in complex color documents

Text extraction in mixed-type documents is a pre-processing and necessary stage for many document applications. In mixed-type color documents, text, drawings and graphics appear with millions of different colors. In many cases, text regions are overlaid onto drawings or graphics. In this paper, a new method to automatically detect and extract text in mixed-type color documents is presented. The proposed method is based on a combination of an adaptive color reduction (ACR) technique and a page layout analysis (PLA) approach. The ACR technique is used to obtain the optimal number of colors and to convert the document into the principal of them. Then, using the principal colors, the document image is split into the separable color plains. Thus, binary images are obtained, each one corresponding to a principal color. The PLA technique is applied independently to each of the color plains and identifies the text regions. A merging procedure is applied in the final stage to merge the text regions derived from the color plains and to produce the final document. Several experimental and comparative results, exhibiting the performance of the proposed technique, are also presented.

Identification of text-only areas in mixed-type documents

Abstract The identification of text areas in a document is crucial for optical character recognition (OCR), imagecompression and image-storage systems. This paper presents a new method for text identification in mixed-type documents. This type of document contains text, drawings and halftones. The proposed method separates the document into text and non-text regions. Thus, the objective is to find, with confidence, the text region of the documents. The method is based on text characteristics such as size, frequency, collinearity and vicinity of connected components, while in the final stage a new texture-analysis technique is applied. For collinearity and vicinity checking, a new technique is used, that overcomes the difficulties of the application of Hough transform. The proposed segmentation method belongs to the bottom-up categories, and is more robust than other techniques. It can identify text regions in difficult cases such as skewed documents, non-rectangular text regions, or text included in drawings or halftone regions. The performance of the method was tested on a variety of images. Its effectiveness is demonstrated by several typical examples.

PLA using RLSA and a neural network

Abstract This paper describes a new method for document page layout analysis. The proposed approach is based on the use of the run-length smoothing algorithm (RLSA) and a neural network block classified (NNBC). The RLSA is used locally and globally for the block segmentation by using optimal pre-estimated smoothing values. The NNBC is used in the classification steps of the method as a tool which classifies the blocks of the document into basic classes or subclasses. The NNBC consists of a principal component analyzer (PCA) and a self-organized feature map (SOFM). The input feature vector is a set of features corresponding to the contents and the relationships of 3×3 masks. This set is selected by using a statistical selection procedure, and provides textural information. In the final step, and after the application of a grouping procedure, the document blocks are classified as text frames and isolated text lines, graphics and halftones, or into secondary subclasses corresponding to special cases of the basic classes. The proposed method can identify blocks that cannot be separated with horizontal and vertical cuts, and gives very correct classification even on documents of bad scanning quality. The performance of the method has been extensively tested on a variety of documents. Several examples illustrate the strength and the effectiveness of the methodology.

Multithresholding of mixed-type documents

Mixed type documents include text, drawings and graphics regions. It is obvious that a technique that can reduce the number of the gray-levels in accordance to the type of each document region could be important for many document applications, such as storage, transmission and recognition. To solve this problem this paper proposes a new method that is called the document multithresholding technique. The method is based on a Page Layout Analysis (PLA) technique and on a neural network multilevel threshold selection approach. In the final document the different block types are stored with the appropriate and limited number of gray-level values. In text and line-drawing blocks, only one threshold is determined whereas in the graphics blocks the optimal number of thresholds is first determined. The performance of the method was extensively tested on a variety of documents.

Locating text in color documents

In complex color documents, text, drawings and graphics occur with millions of different colors. In many cases, text regions are overlaid onto drawings or graphics. A new method is proposed for the automatic detection and extraction of text in mixed type color documents. The proposed method is based on a combination of an adaptive color reduction (ACR) technique and a page layout analysis (PLA) approach. The ACR technique is used to obtain the optimal number of colors. Then, the image is split to separable binary images, each one corresponding to every principal color. The PLA technique is applied independently to each one of the color planes and identifies the text regions. A merging procedure is applied in the final stage to merge the text regions derived from the color planes and to produce the final document.

A New Adaptive Color Quantization Technique

This paper proposes a new algorithm for color quantization. The proposed approach achieves color quantization using an adaptive tree clustering procedure. In each node of the tree a self-organized Neural Network Classifier (NNC) is used which is fed by image color values and additional local spatial features. The NNC consists of a Principal Component Analyzer (PCA) and a Kohonen self-organized feature map (SOFM) neural network. The output neurons of the NNC define the color classes for each node. The final image not only has the dominant image colors, but also its texture approaches the image local characteristics used. For better classification, split and merging conditions are used in order to define if color classes must be split or merged. To speed-up the entire algorithm and reduce memory requirements, a fractal scanning subsampling technique is used.

Document block identification using a neural network

This paper describes a new method that clusters the content of a mixed type document in text or nontext areas. The proposed approach is based on a new set of textural features combined with a two stage neural network classifier. The neural network classifier consists of a principal components analyzer and a Kohonen self organized feature map. Document blocks are classified as text, graphics and halftones or to secondary subclasses corresponding to special cases of the primal classes. The proposed method can identify text regions included in graphics or even overlapped regions, that is, regions that cannot be separated with horizontal and vertical cuts. The performance of the method was extensively tested on a variety of documents with very promising results.