The Study On Background Clutter Metrics Based On Human Vision

With the rapid development of optical imaging technology, the role of background clutter in target acquisition is more and more important. Accurate quantification of background clutter has become a key factor for the performance prediction of optoelectronic imaging system. In this paper, we propose a background clutter metric based on human eyes, which makes the results more accurate.Since the first background clutter metric was put forward, the domestic and foreign scholars have made a lot of efforts in this field, and put forward a variety of clutter metrics. Through reading relevant literature and researching related metrics we found that some of the metrics are too simple and quantitative results are not accurate enough. Therefore, the authors propose a background clutter metric which will be more accurate.In this paper, the following aspects are studied:（1） in human eyes, the structure information of the image is highly adaptive. As background clutter is relative to the target, we can use structural similarity of background and target to measure the intensity of background clutter in the scene. We define the structural information in the image as the feature of the scene, which has nothing to do with the average brightness and contrast. Image brightness is defined as the average gray value, Image contrast is defined as the standard deviation of gray. After being subtracted the mean value of the signal, the signal has a unit standard, and the normalized signal is used as the structure of the signal. Finally, the similarity measure of the target structure is obtained by combining the brightness, contrast and structure of the image.（2） Taking into account the fact that the image edge is the basic feature of the image, Image edge has a great influence on the subjective perception of human eyes, and the change of image edge information means the change of the basic content or structure. Therefore, we should take into account the edge information of the image.（3） In the process of target detection, color information is also a factor affecting the human eye to obtain the target location, so the color information is also considered in the text.（4） Combining the above factors, a new background clutter scale is proposed, which considers many factors, fully embodies the vision of the human eye, and is simple in calculation. Based on the correlation theory and error theory, we use the Search₂ database to verify the proposed clutter metrics, the superiority of the proposedmetrics has been demonstrated by comparing with the existing background clutter metrics.（5） take into account the sensitivity of the human eye to texture. In this paper, we try to use the LBP texture operator in the field of clutter quantization and calculate the differences of the texture features extracted to build a new clutter metric. Then we use the Search₂ database to verify the proposed clutter metric, through comparing with the existing background clutter metrics, we proved that the metric is better than the traditional ones.