An ISP system for low-light image enhancement based on dark channel prior and bilateral filtering
By integrating dark channel prior and bilateral filtering algorithms into an ISP system, and employing fixed-point arithmetic and pipeline architecture optimization, the problems of high computational complexity, color distortion, and low data throughput in low-light image enhancement are solved, achieving efficient image enhancement processing.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- QIANGHUA TIMES (CHENGDU) TECH CO LTD
- Filing Date
- 2026-06-23
- Publication Date
- 2026-07-21
AI Technical Summary
Existing low-light enhancement algorithms suffer from high computational complexity and poor real-time performance when implemented in hardware. Directly applying dark channel priors to low-light scenes can easily lead to color distortion and halo artifacts. The floating-point operation implementation of bilateral filtering is resource-intensive and difficult to integrate. Furthermore, insufficient cross-clock domain synchronization, de-mosaicing, and collaborative design with enhancement modules in the ISP pipeline result in low data throughput and pipeline congestion.
A low-light image enhancement ISP system based on dark channel prior and bilateral filtering is adopted. Through the integrated design of data preprocessing module, demosaic module and low-light enhancement module, a lookup table is used to store the fixed-point ratio of pre-calculated transmittance coefficient to atmospheric light value. Multipliers and dividers are used to realize fixed-point number operations. Reliable data transmission is achieved through flag signal generation, two-stage synchronous trigger synchronization and data sampling mechanism. A 5×5 pixel matrix window is used to construct and pipeline architecture to optimize the data flow between modules.
It significantly reduces hardware implementation costs, improves computational efficiency, suppresses color distortion and halo artifacts, increases data throughput and processing frame rate, and solves the problems of high computational complexity and pipeline blockage in existing technologies.
Smart Images

Figure CN122434802A_ABST