CAJ | 학술논문

为了克服基于小波变换的双阈值递归边缘检测算法检测到的图像边缘不连续的缺点，提出了将自适应梯度算法和双阈值递归相结合的图像边缘检测方法。通过获得二维小波变换的模极大值点来检测图像的候选边缘点；在确定双阈值的情况下，选择高于较大阈值的为边缘，低于较小阈值的为非边缘；通过对候选边缘点的3×3邻域和5×5邻域的跟踪补偿，确定介于双阈值之间的候选边缘点是否为最终的边缘点。实验结果表明，在同样阈值下，该算法比双阈值递归边缘检测算法获得的图像边缘连续性更好，能够更有效地抑制噪声，边缘检测效果更好。
위료극복기우소파변환적쌍역치체귀변연검측산법검측도적도상변연불련속적결점，제출료장자괄응제도산법화쌍역치체귀상결합적도상변연검측방법。통과획득이유소파변환적모겁대치점래검측도상적후선변연점；재학정쌍역치적정황하，선택고우교대역치적위변연，저우교소역치적위비변연；통과대후선변연점적3×3린역화5×5린역적근종보상，학정개우쌍역치지간적후선변연점시부위최종적변연점。실험결과표명，재동양역치하，해산법비쌍역치체귀변연검측산법획득적도상변연련속성경호，능구경유효지억제조성，변연검측효과경호。
To avoid the image edges detected by double threshold recursion algorithm based on wavelet transform were disconti-nuous,a new image edge detection algorithm using adaptive gradient algorithm and appointed double threshold was proposed. First of all,by means of detecting the 2D wavelet transform modulus maxima,the rough edge points were selected.Secondly, the edge points greater than high threshold were final edge points,and the points less than low threshold points were not final edge points.Finally,the edges points between low and high threshold edges were selected from 3 ×3 and 5 ×5 neighborhood points by compensatory method.Experiment results showed that the image edges detected by this method were better than that detected by the double threshold recursion edge detection algorithm in the same threshold.And noise was suppressed effectively by this method,and the edge detection effect was improved greatly.