Texture is one of the most important visual information, which makes the computer recognizes the material of the objects by analyzing the surface texture of them. Therefore, image texture recognition becomes one of the key branches of digital image recognition. Texture recognition has been widely used in the fields of fabric texture automatic detection, medical image analysis and remote sensing image processing, which further proves its research importance and practical value. With comparisons of different methods and the complex degree of function, the thesis finally applies gray level co-occurrence matrix to functionalize the texture recognition. Gray level co-occurrence matrix method is based on the second-order conditional probability density function of the estimated image. Under the condition of Visual C++ compilation, firstly the computer computes the gray level co-occurrence matrix of 4 angles from the gray image. Then, it will compute the probability of every pair pixel in the gray level co-occurrence matrix. Finally, it will compute the 5 Characteristic values according to the formula of energy, entropy, inertia matrix, correlation, and local balance. According to the experiment, this recognized system of that method can precisely compute those five characteristic values leading to the distinguishing of different materials. At the same time, the gray level co-occurrence matrix is proved to be a precisely practical method.