Pear anthracnose and pear black spot are serious diseases that occur during the growth of pears. The symptoms of these two diseases are very similar and it is difficult to distinguish, which leads to the inconvenience of prescribing the right medicine to these two kinds of leaves in actual production. In response to the status quo, taking ‘Dangshan'pear leaves as the study object, the feasibility of using hyperspectral technology to identify anthracnose and black spot on pear leaves was explored. First of all, the hyperspectral imaging system was used to collect the hyperspectral images of the normal leaves, anthracnose leaves and black spot leaves of ‘Dangshan' pear, and extract the average spectral reflectance of the images. The multiplicative scatter correction method (MSC), Savitzky-Golay convolution smoothing method and standard normal variate method (SNV) were used respectively to preprocess the original spectral data. Then the principal component analysis (PCA), successive projections algorithm (SPA), uniformative variable elimination (UVE), competitive adaptive reweighted sampling algorithm (CARS), and shuffled frog leaping algorithm(SFLA) were used to extract characteristic wavelengths, respectively, and totally 27, 12, 15, 26 and 20 characteristic wavelengths were obtained, and using them as input variables for later modeling. After comparison, it was found that in the support vector machine (SVM) classification and recognition model based on characteristic wavelength and the BP neural network classification and recognition model based on characteristic wavelength, the SPA-SVM recognition model had the best effect during all models, the accuracy rate of the model's test set was 93.25%, and the accuracy rate of the model's modeling set was 94.80%. The test results proved that hyperspectral technology can effectively identify the black spot and anthracnose of ‘Dangshan’ pear leaves.