Mastering the tempo-spatial distribution information of enclosure culture areas is useful for a scientific planning of enclosure culture areas of shallow lakes and making effective measurements to improve water quality. This study took Lake Yangcheng as a study area, which is a typical enclosure culture area in the lower reaches of Yangtze River. An extraction algorithm of remote sensing images is proposed through the gradient transformation of remote sensing data based on the differences between enclosure and non-enclosure areas in spectral space. A high-resolution ZY-3 image was used to acquire the spatial distribution of enclosure culture areas in shallow lakes. The enclosure culture areas extracted using the proposed extraction algorithm and the manual visual interpretation were compared to evaluate the classification accuracy. The results show that the overall classification accuracy reached to 90.66%. The proposed method could be used to monitor the dynamic changes of enclosure culture areas in shallow lakes based on the ZY-3 images.