Research Of Color Image Segmentation Based On The Combination Of OTSU And Region Growing Method

Image segmentation is an important research field of information technology processing, is also now technology research hotspot and difficulty. With the rapid development of modern technology, image segmentation technology is making rapid progress as well. However, due to the image acquisition process is influenced by many external factors, such as straight and strong sunshine, which make the image too light or too dark, and sometimes the missing part of image could bring some trouble to the image segmentation and cause technical difficulties. Furthermore, image segmentation research is mainly aimed at one type of image segmentation, there is no general segmentation algorithm. Therefore, for image segmentation, it is of research value and significance to complete the segmentation preprocessing and improve image segmentation algorithm.In this paper, the related image segmentation algorithm is analyzed, and is integrated in the use of experiment on the basis of the theoretical analysis. We focused on the use of OTSU with traditional region growing algorithm for segmentation of color images with complex background first, on the basis of experiment, improved the over segmentation problem of traditional region growing algorithm; and then we studied the segmentation of gray image with multiple target, and to a certain threshold segmentation improvement.Region growing algorithm is one of the image segmentation algorithm used frequently, and has some shortcomings and deficiencies. In the third chapter of this paper, Aiming at the over-segmentation problem of traditional image segmentation based on region growing algorithm with complex background,we propose a new method for segmenting image based on the HIS color model. Firstly,we extract the H component of HIS color model,and apply the OTSU method to perform preliminary segmentation to get the foreground part of the H component on the base of morphology denoising. Secondly,the region growing method is continually used for segmenting the foreground part. Finally,morphology of corrosion method is used to remove the debris. The experimental results show that this algorithm can segment the target image more complete and with better results.Threshold segmentation algorithm is another classic algorithms used in image segmentation. Threshold segmentation has the advantages of quick, convenient. Though the effect of threshold segmentation algorithm for single target image segmentation is quite good, it is easy to cause the wrong division when applied with images with multiple targets, and segmentation speed is quite slow in this case. In this paper, in order to segment multiple targets in the gray image, a segmentation algorithm based on iterative multiple thresholds is proposed. After nonlinear stretching of gray histogram, the two value mask method is used to deal with multiple gray intervals, and then the iterative threshold algorithm is applied to process the targets repeatedly. Experimental results show that this algorithm can effectively reduce the computation complexity compared with that of the single threshold exhaustive search algorithm, and the segmentation results are also satisfactory.