Texture for Script identification

The problem of determining the script and language of a document image has a number of important applications in the field of document analysis, such as indexing and sorting of large collections of such images, or as a precursor to optical character recognition (OCR). In this paper, we investigate the use of texture as a tool for determining the script of a document image, based on the observation that text has a distinct visual texture. An experimental evaluation of a number of commonly used texture features is conducted on a newly created script database, providing a qualitative measure of which features are most appropriate for this task. Strategies for improving classification results in situations with limited training data and multiple font types are also proposed.

Hell-vetica

Helvetica (connotes Swiss typeface) has been used the most widely from street signs to government campaign posters since 1957. Helvetica represents a great leap forward for modernity: clean, sans-serif, optimistic. However in history, there was a movement against Helvetica among American artists and designers since David Carson and Paula Scher indicted Helvetica as the cause of Vietnam war. Paradoxically, we celebrated its 50th birthday in 2007. Helvetica’s message it this: “you are going to get to your destination on time; your plan will not crash; your money is safe in our vault; we will not break the package; the paperwork has been filled in; everything is going to be OK” (Finlo Rohrer, Helvetica At 50, BBC News Magazine 9 May 2007). The artwork, Hell-vetica describes its characteristic of double agent for modernism and postmodernism in this contemporary era by combination of a stylised graphical form of a heart shape in red and a typographical manipulation - Hell-vetica.

Digit recognition using trispectral features

Features derived from the trispectra of DFT magnitude slices are used for multi-font digit recognition. These features are insensitive to translation, rotation, or scaling of the input. They are also robust to noise. Classification accuracy tests were conducted on a common data base of 256× 256 pixel bilevel images of digits in 9 fonts. Randomly rotated and translated noisy versions were used for training and testing. The results indicate that the trispectral features are better than moment invariants and affine moment invariants. They achieve a classification accuracy of 95% compared to about 81% for Hu's (1962) moment invariants and 39% for the Flusser and Suk (1994) affine moment invariants on the same data in the presence of 1% impulse noise using a 1-NN classifier. For comparison, a multilayer perceptron with no normalization for rotations and translations yields 34% accuracy on 16× 16 pixel low-pass filtered and decimated versions of the same data.

Kirsten ras mutations in patients with colorectal cancer : the ‘RASCAL II’ study

Researchers worldwide with information about the Kirsten ras (Ki-ras) tumour genotype and outcome of patients with colorectal cancer were invited to provide that data in a schematized format for inclusion in a collaborative database called RASCAL (The Kirsten ras in-colorectal-cancer collaborative group). Our results from 2721 such patients have been presented previously and for the first time in any common cancer, showed conclusively that different gene mutations have different impacts on outcome, even when the mutations occur at the same site on the genome. To explore the effect of Ki-ras mutations at different stages of colorectal cancer, more patients were recruited to the database, which was reanalysed when information on 4268 patients from 42 centres in 21 countries had been entered. After predetermined exclusion criteria were applied, data on 3439 patients were entered into a multivariate analysis. This found that of the 12 possible mutations on codons 12 and 13 of Kirsten ras, only one mutation on codon 12, glycine to valine, found in 8.6% of all patients, had a statistically significant impact on failure-free survival (P = 0.004, HR 1.3) and overall survival (P = 0.008, HR 1.29). This mutation appeared to have a greater impact on outcome in Dukes’ C cancers (failure-free survival, P = 0.008, HR 1.5; overall survival P = 0.02, HR 1.45) than in Dukes’ B tumours (failure-free survival, P = 0.46, HR 1.12; overall survival P = 0.36, HR 1.15). Ki-ras mutations may occur early in the development of pre-cancerous adenomas in the colon and rectum. However, this collaborative study suggests that not only is the presence of a codon 12 glycine to valine mutation important for cancer progression but also that it may predispose to more aggressive biological behaviour in patients with advanced colorectal cancer.

Text segmentation in unconstrained hand-drawings in whiteboard photos

In this paper we present a robust method to detect handwritten text from unconstrained drawings on normal whiteboards. Unlike printed text on documents, free form handwritten text has no pattern in terms of size, orientation and font and it is often mixed with other drawings such as lines and shapes. Unlike handwritings on paper, handwritings on a normal whiteboard cannot be scanned so the detection has to be based on photos. Our work traces straight edges on photos of the whiteboard and builds graph representation of connected components. We use geometric properties such as edge density, graph density, aspect ratio and neighborhood similarity to differentiate handwritten text from other drawings. The experiment results show that our method achieves satisfactory precision and recall. Furthermore, the method is robust and efficient enough to be deployed in a mobile device. This is an important enabler of business applications that support whiteboard-centric visual meetings in enterprise scenarios. © 2012 IEEE.