ADAPTIVE ILLUMINATION CORRECTION ALGORITHM BASED ON RETINEX TECHNOLOGY FOR CAMERA TRAP IMAGES

Abstract. Camera traps generating a huge number of images help to study and monitor the wildlife. However, camera work at any time day under weather conditions. Therefore, many have low or high illumination, blurring, other defects. This complicates image analysis by both humans computer systems. In this study, we develop an adaptive illumination correction algorithm based on modified Multi-Scale Retinex (MSR). First, accelerate computation using recursive implementation Gaussian filter utilizing look-up tables find logarithms new brightness values. Second, response MSR function is transformed threshold normalization improve quality. The upper lower thresholds are calculated statistical information. Finally, offer automatic adjustment parameters depending area in order increase usability. Proposed was tested with various settings set obtained from traps. Experimental results show potential for its application.