红外与激光工程
紅外與激光工程
홍외여격광공정
INFRARED AND LASER ENGINEERING
2014年
6期
1807-1811
,共5页
张红辉%罗海波%余新荣%丁庆海
張紅輝%囉海波%餘新榮%丁慶海
장홍휘%라해파%여신영%정경해
盲元检测%响应特征%直方图分解
盲元檢測%響應特徵%直方圖分解
맹원검측%향응특정%직방도분해
blind- pixel detection%response characteristic%histograms decomposition
红外探测器受材料和工艺等的限制,普遍存在盲元问题,降低了红外成像系统的图像质量,影响了目标检测系统的检测概率和虚警率,因此,有效检测盲元是红外图像处理领域的一个重要研究方向。在分析了有效像元校正模型的基础上,指出均匀辐照下红外焦平面阵列(IRFPA)有效像元非均匀校正后灰度图像的均值(MEAN)和标准差(STD)具有正态分布特征,对特征直方图采用自适应正交投影分解法进行高斯分解,得到有效像元校正后的MEAN和STD特征分布区间,最终得到盲元的分类准则。对IRFPA进行的验证实验结果表明,该方法获得了令人满意的结果,证实了所提出检测方法的有效性和科学性。
紅外探測器受材料和工藝等的限製,普遍存在盲元問題,降低瞭紅外成像繫統的圖像質量,影響瞭目標檢測繫統的檢測概率和虛警率,因此,有效檢測盲元是紅外圖像處理領域的一箇重要研究方嚮。在分析瞭有效像元校正模型的基礎上,指齣均勻輻照下紅外焦平麵陣列(IRFPA)有效像元非均勻校正後灰度圖像的均值(MEAN)和標準差(STD)具有正態分佈特徵,對特徵直方圖採用自適應正交投影分解法進行高斯分解,得到有效像元校正後的MEAN和STD特徵分佈區間,最終得到盲元的分類準則。對IRFPA進行的驗證實驗結果錶明,該方法穫得瞭令人滿意的結果,證實瞭所提齣檢測方法的有效性和科學性。
홍외탐측기수재료화공예등적한제,보편존재맹원문제,강저료홍외성상계통적도상질량,영향료목표검측계통적검측개솔화허경솔,인차,유효검측맹원시홍외도상처리영역적일개중요연구방향。재분석료유효상원교정모형적기출상,지출균균복조하홍외초평면진렬(IRFPA)유효상원비균균교정후회도도상적균치(MEAN)화표준차(STD)구유정태분포특정,대특정직방도채용자괄응정교투영분해법진행고사분해,득도유효상원교정후적MEAN화STD특정분포구간,최종득도맹원적분류준칙。대IRFPA진행적험증실험결과표명,해방법획득료령인만의적결과,증실료소제출검측방법적유효성화과학성。
Due to the influence of material, fabrication technology and so on, blind- pixels exist on infrared focal plane array detectors. It reduces the infrered image quality and object detection capability, improves the object inveracious rate. Effective blind- pixels detection for infrared image process is an important development issue. Based on the analysis of the model for the effective pixel corrected response, the approach was proposed that, when the corrected response was uniform, the means (MEAN) and the standard deviation (STD) were both normal distributed random variables for all effective pixels. Through adaptive projective decomposition of MEAN and STD histograms, the MEAN and STD distribution range of effective pixel could be effectively estimated, and the detection rule for blind- pixel was got finally. The experimental result by applying real infrared image shows the proposed method for IRFPA blind- pixel detection is scientific and effective.