In this paper a technique of Hough detector threshold procedure for moving target detection in conditions of randomly arriving impulse interference with Poisson distributed flow and Raleigh amplitude distribution is proposed. The expressions of detection and false alarm probability are derived for a highly fluctuating Swerling II target. A comparative analysis of the performance of a Hough detector with fixed threshold and other five Hough detection structures keeping constant false alarm rates is done. These are CA CFAR (Cell Averaging Constant False Alarm Rate), EXC CFAR (Excision Constant False Alarm Rate), CFAR BI (Constant False Alarm Rate with Binary Integration), EXC CFAR BI (Excision Constant False Alarm Rate with Binary Integration) and API CFAR (Adaptive censoring Post detection Integration Constant False Alarm Rate). A method for losses estimation, which allows choosing of optimal detector parameters, is developed. Obtained are estimates of the effectiveness of Hough detector in presence of randomly arriving impulse interference and they are compared to patterns researched from other authors. The achieved results can be successfully applied for radar target detection and in the existing communication network receivers that use pulse signals.