CAJ | 학술논문

为实现多种复杂海天背景下红外舰船目标的有效检测，提出了一种基于局部边缘梯度特征分析的舰船目标检测方法。利用红外舰船目标图像的特征，改进自适应 Butterworth 高通滤波对图像进行预处理。通过计算局部边缘梯度特征和聚类分析确定初始目标区域，比较梯度特征参数并进行多尺度搜索准确检测出舰船目标。对多种复杂海天场景红外图像进行仿真实验，结果表明，该算法能够有效准确地检测出多种复杂海天场景下的红外舰船目标，且算法简单，易于实现，适应面广。
위실현다충복잡해천배경하홍외함선목표적유효검측，제출료일충기우국부변연제도특정분석적함선목표검측방법。이용홍외함선목표도상적특정，개진자괄응 Butterworth 고통려파대도상진행예처리。통과계산국부변연제도특정화취류분석학정초시목표구역，비교제도특정삼수병진행다척도수색준학검측출함선목표。대다충복잡해천장경홍외도상진행방진실험，결과표명，해산법능구유효준학지검측출다충복잡해천장경하적홍외함선목표，차산법간단，역우실현，괄응면엄。
In order to fulfill the warship target extraction of multiple complex scenes, this paper presents an algorithm based on gradient characteristic of infrared complicated sea-sky image. The infrared image was preprocessed by improved Butterworth high-pass filter which using features of infrared sea-sky image. Then the initial target’s regional can be obtained by clustering the peak value of edge gradient. The correct warship was extracted by comparing the edge gradient parameters and multi-scale search. Experimental results indicated that this algorithm can detect the warship under various complicated sea-sky background effectively, and have good adaption, also can be implemented easily.