摘要
为了校正形变像素,提升图像质量,提出一种基于近红外成像光谱的人体动态运动点像素校正系统,通过探究光谱的成像原理与像素发生形变的原因对其展开研究,利用三帧连续图像的配准与差分相乘法对人体动态轮廓进行获取,经过形体学与形态学膨胀处理后,得到运动点像素及其空间域的相关性,统计后的像素数据会被传至整体统计模块中,从而获得通道统计均值,将整体统计模块聚合后解得整体统计值,根据像素增益标量与暗补偿和光强响应,采用最小平方法拟合理想矩阵与实际矩阵,再基于方位向脉冲压缩运动点与距离像素形变量表达式,对形变像素完成校正,最后运用中值滤波对校正后图像实施噪声点去除。仿真实验表明,系统在校正形变像素时效果明显且具有极高的准确性,可使图像细节得到精准还原,大幅度提升图像质量。
In order to correct the deformed pixels and improve the image quality,this paper proposes a pixel correction system of human body dynamic motion points based on near-infrared imaging spectrum.By exploring the imaging principle of the spectrum and the causes of pixel deformation,we use the registration and difference multiplication of three consecutive images to obtain the dynamic contour of the human body,which is processed by the physique and morphological expansion.After that,the correlation of moving point pixels and their spatial domain is obtained,and the statistical pixel data will be transmitted to the overall statistical module to obtain the channel statistical mean value.The overall statistical module is aggregated to get the overall statistical value.According to the pixel gain scalar and dark compensation and light intensity response,the least square method is used to fit the ideal matrix and the actual matrix and then based on the azimuth pulse compression transport.The shape variable expressions of moving point and distance pixels are used to correct the deformed pixels.Finally,the median filter is used to remove the noise points of the corrected image.The simulation results show that the system has the apparent effect and high accuracy when correcting orthomorphic variable pixels,making the image details accurately restored and significantly improving image quality.
作者
张静
周健
ZHANG Jing;ZHOU Jian(Three Gorges University,Yichang Hubei 443000,China)
出处
《激光杂志》
CAS
北大核心
2021年第7期60-64,共5页
Laser Journal
基金
国家自然科学基金(No.11201385)。
关键词
近红外成像光谱
动态运动点
像素校正
距离单元
near-infrared imaging spectrum
dynamic moving point
pixel correction
distance unit
