We explore the effect of dimensionality on the nearest neighbor problem. We show that under a broad set of conditions (much broader than independent and identically distributed dimensions), as dimensionality increases, the distance to the nearest data point approaches the distance to the farthest data point. To provide a practical perspective, we present empirical results on both real and synthetic data sets that demonstrate that this effect can occur for as few as 10-15 dimensions. These results should not be interpreted to mean that high-dimensional indexing is never meaningful; we illustrate this point by identifying some high-dimensional workloads for which this effect does not occur. However, our results do emphasize that the methodology used almost universally in the database literature to evaluate high-dimensional indexing techniques is flawed, and should be modified. In particular, most such techniques proposed in the literature are not evaluated versus simple linear scan, and are evaluated over workloads for which nearest neighbor is not meaningful. Often, even the reported experiments, when analyzed carefully, show that linear scan would outperform the techniques being proposed on the workloads studied in high (10-15) dimensionality!.

When is nearest neighbor meaningful

Optical character recognition (OCR) is an important research area in the field of pattern recognition. A lot of research has been done on OCR in the last 60 years. There is a large volume of paper-based data in various libraries and offices. Also, there is a wealth of knowledge in the form of ancient text documents. It is a challenge to maintain and search from this paper-based data. At many places, efforts are being done to digitize this data. Paper based documents are scanned to digitize data but scanned data is in pictorial form. It cannot be recognized by computers because computers can understand standard alphanumeric characters as ASCII or some other codes. Therefore, alphanumeric information must be retrieved from scanned images. Optical character recognition system allows us to convert a document into electronic text, which can be used for edit, search, etc. operations. OCR system is the machine replication of human reading and has been the subject of intensive research for more than six decades. This paper presents a comprehensive survey of the work done in the various phases of an OCR with special focus on the OCR for ancient text documents. This paper will help the novice researchers by providing a comprehensive study of the various phases, namely, segmentation, feature extraction and classification techniques required for an OCR system especially for ancient documents. It has been observed that there is a limited work is done for the recognition of ancient documents especially for Devanagari script. This article also presents future directions for the upcoming researchers in the field of ancient text recognition.

Ancient text recognition: a review

this paper a support vector machine based handwritten numerals recognition system of Manipuri Script (Meetei Mayek) is investigated. We have used various feature extraction technique such as background directional distribution (BDD), zone based diagonal, projection histograms and Histogram Oriented Gradient features. In Background Directional Distribution (BDD) features background distribution of neighboring background pixels to foreground pixels in 8-different directions is considered forming a total of 128 features. For the computation of diagonal features, the whole image is divided into 64 zones of equal dimension each of size 4×4 pixels then features are extracted from the pixels of each zone by moving along the diagonal, thus consisting of 64 features in total. Projection Histograms count the number of foreground pixels in different directions such as vertical, horizontal, horizontal, left diagonal and right diagonal creating a total of 190 features. The HOG based feature is computed over the validation data set, was achieved by means of 9 rectangular cells and 9 bin histogram per cell Different combinations of these features are used for forming various feature vectors. These feature vectors are classified by using SVM classifier as 5-fold cross validation with RBF (radial basis function) kernel. Experimental results show that the proposed system performs well with the combined features and is robust to the writing variations that exist between persons and for a single person at different instances, thus being promising for user independent recognition of Meetei Mayek numeral.

/pdf/recognition-of-handwritten-numerals-of-manipuri-script-2el7qm23zs.pdf

Recognition of Handwritten Numerals of Manipuri Script

In Cursive languages like Urdu, segmentation of handwritten text lines is such a problem because of context sensitivity, diagonality of text etc. In this work, we presented a simple and robust line segmentation algorithm for Urdu handwritten and printed text. In the proposed line segmentation algorithm, modified header and baseline detection method are used. This technique purely depends on the counting pixels approach. Which efficiently segment Urdu handwritten and printed text lines along with skew detection. Handwritten and printed Urdu text dataset is manually generated for evaluating algorithm. Dataset consists of 80 pages having 687 handwritten Urdu text lines and printed dataset consist of 48 pages having 495 printed text lines. The algorithm performed significantly well on printed documents and handwritten Urdu text documents with well-separated lines and moderately well on a document containing overlapping words.

An Efficient Segmentation Technique for Urdu Optical Character Recognizer (OCR)

This paper proposes a novel algorithm for the vectorization of ordered sets of points, named Fast Total Least Squares (FTLS) vectorization. The emphasis was put on low computational complexity, which allows it to be run online on a mobile device at a speed comparable to the fastest algorithms, such as the Douglas–Peucker (DP) algorithm, while maintaining a much higher quality of the approximation. Our approach is based on the total least squares method, therefore all the points from the cloud contribute to its approximation. This leads to better utilization of the information contained in the point cloud, compared to those algorithms based on point elimination, such as DP. Several experiments and performance comparisons are presented to demonstrate the most important attributes of the FTLS algorithm.

Fast total least squares vectorization

This paper presents an approach for text line segmentation which combines connected component based and projection based information to take advantage of aspects of both methods. The proposed system finds baselines of each connected component. Lines are detected by grouping baselines of connected components belonging to each line by projection information. Components are assigned to lines according to different distance metrics with respect to their size. This study is one of the rare studies that apply line segmentation to Ottoman documents. Further, it proposes a new method, Fourier curve fitting, to detect the peaks in a projection profile. The algorithm is demonstrated on different printed and handwritten Ottoman datasets. Results show that the method manages to segment lines both from printed and handwritten documents under different writing conditions at least with 92% accuracy.

/pdf/a-hybrid-for-line-segmentation-in-handwritten-documents-2ksrqjjl9x.pdf

A Hybrid for Line Segmentation in Handwritten Documents

In this study, we address the problem of matching patterns in Kufic calligraphy images. Being used as a decorative element, Kufic images have been designed in a way that makes it difficult to be read by non-experts. Therefore, available methods for handwriting recognition are not easily applicable to the recognition of Kufic patterns. In this study, we propose two new methods for Kufic pattern matching. The first method approximates the contours of connected components into lines and then utilizes chain code representation. Sequence matching techniques with a penalty for gaps are exploited for handling the variations between different instances of sub-patterns. In the second method, skeletons of connected components are represented as a graph where junction and end points are considered as nodes. Graph isomorphism techniques are then relaxed for partial graph matching. Methods are evaluated over a collection of 270 square Kufic images with 8,941 sub-patterns. Experimental results indicate that, besides retrieval and indexing of known patterns, our method also allows the discovery of new patterns.

/pdf/matching-islamic-patterns-in-kufic-images-4cf0vxlbp2.pdf

Matching Islamic patterns in Kufic images

Analyzing and interpreting human actions is an important and challenging area of computer vision. Different solutions are used for representing human actions; we prefer to use spatio-temporal interest points for motion descriptors. Besides, the space-time interest point feature space is considerably high-dimensional and it is hard to eliminate the curse of dimensionality with traditional similarity functions. We apply a matching based approach for high dimensional feature space that matches sequences to classify actions.

/pdf/action-recognition-in-a-high-dimensional-feature-space-3iy0veh83l.pdf

Action recognition in a high-dimensional feature space

In this paper, as a first step to an easy and convenient way to access the manuscripts of Ataturk with a word based search engine, the preprocessing of digitalized documents and their line and word segmentation is studied. The techniques that are applied on printed documents may not yield satisfactory results. Due to this fact, more developed techniques are decided to be applied consisting of a technique based on Hough transform [1] for line segmentation and a technique that is based on dealing with skewness of lines for word segmentation. The results, which are acquired through studies that are conducted on the documents provided by Afet Đnan and consisting of 30 pages [2], prove to be highly accurate and promising for future researches.

Processing the manuscripts of Atatürk

Within Ottoman Text Archive Project a web interface to aid in uploading, binarization, line and word segmentation, labeling, recognition and testing of the Ottoman Turkish texts has been developed. It became possible to retrieve expert knowledge of scholars working with Ottoman archives through this interface, and apply this knowledge in developing further technologies in transliteration of historical manuscripts.

Hande Adiguzel

Papers

A Hybrid for Line Segmentation in Handwritten Documents

Matching Islamic patterns in Kufic images

Action recognition in a high-dimensional feature space

Processing the manuscripts of Atatürk

OTAP ottoman archives internet interface