A hardware architecture that implements an adaptive filter based on energy analysis of radar echoes to improve the detection of the Constant False Alarm Rate (CFAR) algorithm is presented. Signal processing based on energy analysis emphasizes the edge of the echoes improving the performance of the detection process. The energy filter coefficients and CFAR parameters are calculated adaptively by the architecture, reconfiguring the block of coefficient weights according to environment conditions. The architecture accelerates the data processing by a pipeline structure and sliding window for the coefficients convolution with data, resulting in high performance operation. Results of implementing the architecture in a FPGA device are presented and discussed.