No abstract

Many techniques have been reported for handwriting-based writer identi"cation. The majority of techniques assume that the written text is "xed (e.g., in signature veri"cation). In this paper we attempt to eliminate this assumption by presenting a novel algorithm for automatic text-independent writer identi"cation. Given that the handwriting of di!erent people is often visually distinctive, we take a global approach based on texture analysis, where each writer's handwriting is regarded as a di!erent texture. In principle, this allows us to apply any standard texture recognition algorithm for the task (e.g., the multi-channel Gabor "ltering technique). Results of 96.0% accuracy on the classi"cation of 1000 test documents from 40 writers are very promising. The method is shown to be robust to noise and contents.

No abstract

Many techniques have been reported for handwriting-based writer identification. Most such techniques assume that the written text is fixed (e.g., in signature verification). In this paper we attempt to eliminate this assumption by presenting a novel algorithm for automatic text-independent writer identification from non-uniformly skewed handwriting images. Given that the handwriting of different people is often visually distinctive, we take a global approach based on texture analysis, where each writers' handwriting is regarded as a different texture. In principle this allows us to apply any standard texture recognition algorithm for the task (e.g., the multi-channel Gabor filtering technique). Results of 96.0% accuracy on the classification of 150 test documents from 10 writers are very promising. The method is shown to be robust to noise and contents.

Coregistration of optical and radar imageries is a major pre-processing step in many remote sensing applications including change detection and interferometric processing. Specially, the coregistration faced more difficulties in radar imageries due to the high noise level and intense spatial-temporal decorrelation. Hence, non-automatic coregistration methods are much more time consuming and inefficient. Because, the probable geometrical and temporal (signal-target interaction) differences between two acquisitions (i.e, master and slave) caused spectral shifts between corresponding pixels. In this case, using grid points as a tie in the coregistration process drastically reduces the reliability of the process while use of the corner points in the images could solve the problem. To resolve this problem, an automatic algorithm based on angular histogram was proposed for coregistration of Synthetic Aperture Radar (SAR) imageries that have taken in different height, time and situation. This method automatically extracted the corners from the slave and master images and used these points as nodal points. However, unlike conventional methods, there is no need to perform the image matching process. The coregistration RMSE value for three case studies related to the Radarsat-2 imageries of Batala area, India, Radarsat-2 imageries of Sendai region, Japan and TerraSAR-X imageries in Sendai region, were 0.29, 0.35 and 0.43, respectively. Comparison of the proposed coregistration method with cubic convolution (CCI) interpolation based coregistration method showed that the accuracy of the proposed method was improved to 5%, 5% and 9% for three datasets, respectively. The results indicated high efficiency and accuracy of the proposed algorithm. Moreover, the proposed algorithm illustrates the high performance in the case that the angle and scale between the master and slave images are relatively large.