A recognition method of mature litchi in natural scene was researched. Through analyzing the gray and color features of litchi, the YCbCr color model was selected, and by using threshold segmentation, the Cr component image was selected to remove the complex background, then morphology operation and connected region label were used to remove random noise, and then the one-dimensional random signal histogram analysis and the fuzzy C-means clustering (FCM) segmentation were used to achieve the recognition of litchi fruit and peduncle. The recognition results under different natural conditions showed that the algorithm could segment the fruit successfully and the recognition rate reached to 95.5%.