In order to solve the problem of hickory nuts’ material mixing which was difficult to screen after breaking the shells, an image processing algorithm for hickory nuts color and texture selection was proposed based on fuzzy c-means clustering algorithm. The program with user interface for the separation of hickory nuts’ shells and kernels was developed on the Labview software platform. By extracting the characteristic data the hickory nuts with separation substances value of hue, saturation, value, angular second moment, entropy, inverse different moment and correlation, building a sample set of characteristic data, optimal data subset is preferred in accordance with the principal component analysis. After using fuzzy clustering algorithm to calculate the cluster center of corresponding object feature set, the membership degree of the test sample to the corresponding cluster center was calculated. According to the principle of maximum degree of membership, the rational classification of the shells, inner septa（including the inner wall of the shell）and kernels from hickory nuts was realized and the kernels were successfully separated from the breakages. As an experimental result of the separation of shells and kernels, this system could correctly classify the test samples and the correct rate of sorting can reach more than 83%. This study provided a reference for the extensive research on testing the color selection and separation of nuts shelled.