This paper proposes a fast look-up table access algorithm for H.265/high efficiency video coding (HEVC). The proposed algorithm incorporates the coding bandwidth, coding bit rate and coding distortion of H.265/HEVC coding unit (CU) to achieve effective coding optimization of both coding bandwidth and coding quality. This proposed design module is intended for fast lookup of optimized H.265/HEVC coding search-range (SR) data at different resolutions generated by a machine learning (ML) method. The optimized SR data of fast lookup table not only optimizes bit misalignment but also effectively utilizes encoding resources and reduces search time for H.265/HEVC motion estimation (ME). As for software implementation, based on the analyses of different algorithms and data structures we choose the red–black tree as the storage structure for data classification, which can better analyze which kinds of search scopes will be used.