Fiber detection method and system

The invention discloses a fiber detection method. The fiber detection method comprises the following steps: in multiple existing processes, determining a master process and sub processes, sending text information corresponding to to-be-processed pictures to each sub process by use of an MPI, determining identification numbers of pictures needing to be processed by each sub process, and triggering the sub processes to process the pictures corresponding to the identification numbers; when the sub processes complete processing of the pictures, indicating that the sub processes are at idle states, sending picture processing results to the master process, determining whether the master process comprises unprocessed to-be-processed pictures, if so, selecting the identification number of one unprocessed to-be-processed picture, and sending the identification number to the sub processes at the idle states by use of the MPI; and when the master process receives processing results of all the pictures, according to the processing results of all the pictures, calculating proportions of cotton fibers and fibrilia in the pictures. By use of such a master-slave parallel processing mode, the problem of consuming too much time in fiber detection in the prior art is solved.