A Camouflage Effect Detection Model for Fixed Targets

The traditional camouflage effect detection mainly implements the evaluation process for a single image, and cannot effectively reflect the statistical characteristics of the target. In order to better simulate the dynamic interpretation process of reconnaissance personnel on the target, a dynamic and statistically characteristic camouflage effect evaluation model is proposed for the problem of camouflage effect detection of fixed targets. Combined with the Mean shift target tracking technology, the model statistically correlates the target with the background eight-link domain and establishes a normalized joint Gaussian distribution. The target's camouflage effect is then evaluated using the distribution of probability density. The experiment performs complete camouflage, partial camouflage and non-disguise on the exit target of a simulated cavern, and calculates the logarithmic amplification probability and statistics of the curve after collecting the data. According to the 3σ criterion, the mean value is compared with a preset threshold value, and corresponds to the original camouflage state, the better camouflage state, and the invalid camouflage state, respectively. The experimental results show that the model can clearly distinguish the different camouflage states.