CAJ | 학술논문

为了满足自动对焦系统的实时性与抗噪性要求，提出了一种基于SUSAN算子的清晰度评价函数。该函数利用SU-SAN边缘提取算子的算法简单、准确度高、抗噪性强的特点，对SUSAN算子边缘检测函数进行改进，将边缘点的USAN值的平方和作为清晰度评价值，并将其运用到自动对焦算法中。将该函数与几个经典的清晰度评价函数进行性能比较，实验结果表明：对于引入噪声前后的图像序列，基于SUSAN算子的清晰度评价函数均具有良好的单峰性、无偏性和较高的灵敏度；对于256×256的对焦窗口图片，该函数在TMS320 C6416硬件平台上的运行时间仅为16 ms。该函数能够满足清晰度评价函数的单峰性、无偏性、高灵敏度等基本特性，同时具有良好的实时性与抗噪性。
위료만족자동대초계통적실시성여항조성요구，제출료일충기우SUSAN산자적청석도평개함수。해함수이용SU-SAN변연제취산자적산법간단、준학도고、항조성강적특점，대SUSAN산자변연검측함수진행개진，장변연점적USAN치적평방화작위청석도평개치，병장기운용도자동대초산법중。장해함수여궤개경전적청석도평개함수진행성능비교，실험결과표명：대우인입조성전후적도상서렬，기우SUSAN산자적청석도평개함수균구유량호적단봉성、무편성화교고적령민도；대우256×256적대초창구도편，해함수재TMS320 C6416경건평태상적운행시간부위16 ms。해함수능구만족청석도평개함수적단봉성、무편성、고령민도등기본특성，동시구유량호적실시성여항조성。
In order to meet the real-time and anti-noise requirements of auto-focusing system , a definition e-valuation function based on SUSAN is proposed .This function is derived from the SUSAN edge detection algo-rithm, which makes the sum of squared edge point′s USAN values as the definition evaluation value , and ap-plies to the auto-focusing system , making use of the high accuracy and strong anti-noise characteristics of SU-SAN algorithm.The definition evaluation function performance is compared between SUSAN function and the other classical functions .Experiment results show that the SUSAN definition evaluation function has the char -acteristics of clear single apex , good unbiased , high sensitivity for the image sequences before and after the in-troduction of noise.In addition, the running time of the function on the TMS 320C6416 hardware platform is 15 ms.It can satisfy the definition evaluation function requirements of single apex , unbiased and sensitivity char-acteristics , as well as strong real time and anti-noise abilities .