CAJ | 학술논문

脉冲耦合神经网络（ PCNN）模型在数字图像分割中得到了广泛的应用，但对网络参数的确定以及最优结果的选则一直是一个难题，主要是以人工经验为主，虽然提出了一些参数自动设置的算法，但都缺乏对模型本身数学理论的研究和分析。通过对PCNN网络神经元点火特性的分析，提出了一种基于图像最大灰度值最小时刻点火的参数自适应设定算法，并针对该算法构造了一种新的用于最优结果选择的判定准则。将其用于lena等图像分割中，取得了与主观评价相一致的结果，而且表现出更快的速度和较好的鲁棒性。
맥충우합신경망락（ PCNN）모형재수자도상분할중득도료엄범적응용，단대망락삼수적학정이급최우결과적선칙일직시일개난제，주요시이인공경험위주，수연제출료일사삼수자동설치적산법，단도결핍대모형본신수학이론적연구화분석。통과대PCNN망락신경원점화특성적분석，제출료일충기우도상최대회도치최소시각점화적삼수자괄응설정산법，병침대해산법구조료일충신적용우최우결과선택적판정준칙。장기용우lena등도상분할중，취득료여주관평개상일치적결과，이차표현출경쾌적속도화교호적로봉성。
Pulse Coupled Neural Network ( PCNN) model has been widely used in digital image processing , but there is still a problem for the determination of network parameters which at present is mostly determined manual -ly.Some self-adaptive algorithms are available , but the research and analysis on the model theory are still less . This paper proposes a method of automatic setting of parameters based on the maximum gray value firing at the minimum iteration times by analyzing the neuron firing characteristics of PCNN , and a new comprehensive evalu-ation criterion of selecting the optimal result is constructed on the basis of this method .It is used in Lena image , and the results are consistent with subjective assessment , which has quicker speed and better robust .