Chlorophyll content is an important biochemical indicator for evaluating plant health status. Therefore, rapid and accurate monitoring of crop chlorophyll content is the key to achieving precise monitoring of crop growth and health. Hyperspectral image technology can non-destructive invert leaf chlorophyll. Using measured hyperspectral data and corn leaf chlorophyll data content as raw data, regular autoencoders are used to reduce the dimensionality and capture the most significant features in the data, so that the reduced data can replace the original data for training. A two-dimensional convolutional network model based on GASF is constructed to invert the chlorophyll value of corn leaves, And compare it with other data processing methods. The results show that compared with other data processing methods, the two-dimensional inversion prediction model using GASF is the best. The Coefficient of determination in the test set is above 0.919, and the Root-mean-square deviation RMSE is below 1.85. Among them, the DAE-GASF-2DCNN prediction result is the best. The Coefficient of determination in the test set is 0.928, and the Root-mean-square deviation RMSE is 1.1619.