The design of FIR filter in veterinary B-ultrasound system FPGA has the following requirements: First of all, the filtering speed of the filter should be considered, because the veterinary B-ultrasound diagnostic instrument has high requirements for real-time performance and collects a large number of high-speed signals, so the filter is also required to filter out the clutter in the echo signal quickly; secondly, the resources required for the filter designed and implemented in FPGA should be small.

The coefficients of the FIR filter designed in the veterinary B-ultrasound system FPGA are generated by MATLAB, so the design of the filter can be regarded as the design of a constant coefficient FIR filter. There are usually five methods for implementing the design of a constant coefficient FIR filter in FPGA: direct method, transposition structure method, traditional structure DA method, simplified structure DA method, and parallel structure DA method.

When the veterinary B-ultrasound system uses the direct method to design FIR filters, although it is intuitive and easy to understand, it has the disadvantages of occupying too many FPGA resources and slow data processing speed; the transposed structure method is a method improved from the direct method. Although its filtering speed and required resources are greatly improved compared with the direct method, its required resources are still slow and the filtering speed is also slow; the speed of the filter designed by various DA methods has been greatly improved, and the design scale has been greatly simplified. Among them, the filter filtering speed implemented by the parallel structure DA method is the fastest, and the filter required by the simplified structure DA method is the least.

In the actual design of the veterinary B-ultrasound machine, because the veterinary B-ultrasound diagnostic instrument has high requirements for real-time performance, the filtering speed of the filter is the most important design indicator. Among these five design methods, the filter designed by the parallel structure DA method has the fastest filtering speed, so the parallel structure DA method is selected to design the FIR filter.