Segmentation Technique Of Arabic Numerals In The Complex Background

The present recognition technology of Arabic numerals in the complex background still can not meet the needs of practical applications. The Segmentation Technique of Arabic numerals in the complex background hinder the development of identification technologies. To rule out the complex background noise interference, and correctly distinguish between noise and the broken strokes of character, achieve the correct characters segmentation, this paper thinking over character segmentation reference to contour extraction , proposed an improved active contour algorithm. The main research contents and innovation achievements are described as follows:Firstly, according to the summary and analysis about the current methods of contour extraction, proposed a new idea that using the active contour method to achieve the broken character extraction.Secondly, the traditional active contour algorithm can not converge in the smooth large depression and can be excessive convergence in the large bump. So, according to structural characteristics of Arabic numerals, in the active contour model, introduce a double heart external energy function, to overcome the deficiency of the traditional active contour algorithm for contour extraction of Arabic numerals.Then, the images in the practical application often contain a variety of noise which impact segmentation and identification, and the traditional active contour algorithm with the image force energy function susceptible to noise impact contours converge, so introduces the energy function of the proportion of the foreground points in the neighborhood instead of the traditional image force energy function to remove noise effects.Furthermore, the paper presents a new algorithm to remove the jagged contours and acquire the smooth detail of the contours.Finally, based on accurate positioning, using of double centric energy function implemented the within contour extraction of the digitalExperimental verification and engineering applications show that, the contour extraction of Arabic numerals used with the active contour algorithm which consider double heart as external energy was better than traditional algorithms. Compared with other character segmentation techniques, Improved Algorithm successfully achieves the correct character segmentation of the broken stroke, has a stronger anti-noise capability, and meets the practical application level.