Rotation and Scale Invariant Feature Extraction Using Complex Zernike Moments Forfarsiand Arabic Handwriting Character

Analyzing Farsi and Arabic handwritten documents is one area in image processing whose target is to transform picture documents into symbolic form. This transformation is conducted o make rapid and easy saving, improvements, retrieval, reuse, searching and transferring documents. Analyzing documents is performed in five stages: pre-processing, segmentation representation, recognition and post-processing. In this research, in first stage the required pre-processing is performed to normalize the image. In recognition stage, zernik moments–based method has been introduced to extract feature of Farsi and Arabic handwritten characters. Outputs of zernik moments are put in systematic clustering to decide about characters. The obtained results show that feature extraction using zernik moments is a suitable method that deals with few rotation-independent featuresand this leads to reduce calculations and increase speed of recognition and stability against rotation. Size–independence is obtained using difference and at least distance of image calculations, because of preprocessing stage which has been performed in this algorithm is transfer-independent. Valid rank of zenik moments to extract features during conducted test is 4 – 38. Cluster applying has led to reduce algorithm expenditure tplog(n) and this one advantage of the suggested algorithm .