The present application relates to a convolution operation optimization method and system, a terminal, and a storage medium. The method comprises: inputting image data in a data memory module into a multithread data cache module, and recording the data characteristics of the image data in each thread; when all threads are filled with the image data, separately performing spatial-temporal similarity analysis on the data characteristics of at least two adjacent threads, and when the data characteristics of the at least two adjacent threads have spatial-temporal similarity, filtering out the image data of at least one of the at least two adjacent threads, using the thread subjected to image data filtration as an idle thread, re-caching the image data inputted by the data memory module, and when all threads are filled with the image data again, performing spatial-temporal similarity analysis again; and performing a convolution operation according to the cached image data, and outputting new image data. The present application greatly reduces the actual amount of the convolution operation, improves data reusability, reduces the overall calculation time of a network, and improves the performance of a chip.