In order to obtain high-precision fiber identification and counting of fibers by color image analysis and clustering features of the experiment, obtained in the YUV space using only U, V components of the image color clustering. Using k-means clustering algorithm for color image clustering, each pixer according to the image clustering properties of the image segmentation.Identification of the fiber count links, using the segmented Hough transform the way out of the fiber image recognition and counting. Experiments show that the algorithm to identify high precision, can adapt to a range of optical changes in the environment, better robustness.