Abstract:To address the issue of distinguishing seed maize from grain maize with remote sensing, a method of multi-temporal OLI/Landsat-8 remote sensing images combined with GeoEye-1 high-resolution texture was proposed. Utilizing the phenological phase differences of all classes from multi-temporal OLI/Landsat-8 images, the C5.0 decision tree classification algorithm was applied to the constructed EVI time-series. According to the texture difference between seed maize and grain maize, thresholds were set to identify seed maize by using GeoEye-1 high-resolution texture information. Finally, Linze County of Zhangye City in Gansu Province was taken as a study area to test the method. The results showed that the overall classification accuracy of multi-temporal OLI/Landsat-8 was 86.31% and the Kappa coefficient was 0.81, the user accuracy of maize identification was 88.39% and the mapping accuracy was 95.35%, which can meet the demand of further identification of seed maize. In contrast, when combined with texture information from high-resolution images, the user accuracy of seed maize was 86.37% and the mapping accuracy was 83.02%, which were higher than those of exclusive OLI/Landsat-8 data source. The conclusion is, this method can play a technical role in monitoring seed field over large range fast and accurately with remote sensing technology, enforcing seed market supervision and improving the authorities’ response time to the market.