Research on quality assessment for gray night vision fusion image of target detection-oriented vision task

To realize the objective measure of the “target detection perceptual quality” index which is proposed for specific visual tasks, a new fusion image target detection objective assessment model is provided based on target neighborhood contrast. In the model design procedure, the eye-sensitive target and background regions are acquired by the segmentation of region of interested with the help of infrared images. Meanwhile, the target neighborhood area is acquired by the analysis on human visual system characteristics and the observing condition of the subjective assessment experiment. Furthermore, the eye-sensitive weight is acquired based on human visual system dimensional integral characteristics. Finally, the multi-target weight is acquired by the human visual system luminance masking effect. Compared to target and background contrast model based on Weber law, the method is better prominently at the accuracy, consistency and monotonicity which has the realistic significance to the practical image fusion system.