Abstract: Timely and accurate information on geospatial distribution of crops in a growing season is critical for monitoring crop growth and forecasting the agricultural disaster and crop yield. Most remote sensing mapping methods were currently covered the crops images in the entire growing season. There was a time lag fail to meet the harsh requirement in the actual agricultural production. Moreover, the relatively generalization ability in time and space can make the crops mapping difficult to identify in most current approaches. The manual debugging in image acquisition has indicated the heavy workload and strong subjectivity, and thereby seriously prevents the automation of crop extraction in the threshold method. The changes of phenological characteristics in winter wheat gradually accumulate among years, and the time generalization ability of threshold can be mainly affected, with the continuous change of global climate. This study aims to propose a novel crops mapping method to rapidly acquire the specific knowledge in season, using the pixel-wise phenological curves. The reflectance imaging data in the growth period of 250 m winter wheat was used to construct the MODIS-NDVI time series data in 2017-2019. A Savizky-Glolay filter and moving average with dynamic step sizes filtering were used to reconstruct the time series datasets. The images of winter wheat during the stopping and jointing stage from Landsat-8 satellite were superimposed, and then the samples were selected to input the Support Vector Machine (SVM) for the extraction of wheat spatial distribution, serving as the prior knowledge and validation data. A Pixel-Matched Model (PMM) was designed to systematically extract the spatial distribution of winter wheat. The pixel similarity was calculated to compare between the prior winter wheat and current corresponding pixels, as well as the prior non-winter wheat pixels and standard curves. A standard deviation iteration method was used to optimize the threshold, in order to efficiently extract the spatial distribution of winter wheat in the growth period. The results showed that the PMM can fully utilize the change of crop phenological characteristics, and thereby to achieve high-precision extraction within the second month after sowing winter wheat. The obtained distribution from the SVM was set as the reference standard, where the overall accuracy of PMM extraction results was 95.49% and F1 score of 0.83. The high precision remained stable, with the deepening of winter wheat growth period, and the increasing of remote sensing data. The PMM can achieve the winter wheat distribution faster and well in large scale, compared with the Reference Curve Model (RCM), sharing the common phenological curve to extract crop distribution. The most advantage of PMM method was to significantly weaken the influence of phenological spatial difference of winter wheat planting, when extracting through the inter-annual comparison of the pixel own curve. The PMM can complete the identification of winter wheat across the years, indicating the strong time generalization ability to realize the automatic inference of winter wheat distribution using historical models. This approach can remarkably reduce the difficulty of extraction, while save the manpower investment in the extent extraction of winter wheat in growing season.