








摘 "要:為實現(xiàn)數(shù)字圖像自適應(yīng)去噪,提出一種基于遺傳蟻群算法(GACA)優(yōu)化的脈沖耦合神經(jīng)網(wǎng)絡(luò)(PCNN)改進(jìn)中值濾波混合圖像去噪方法(GACA-PCNN-MF)。通過將遺傳算法(GA)和蟻群算法(ACO)相結(jié)合使GA的計算結(jié)果用于增強ACO早期信息素,最終使ACO在正反饋機制中加速優(yōu)化PCNN關(guān)鍵參數(shù),然后使用優(yōu)化后的PCNN改進(jìn)中值濾波技術(shù)進(jìn)行圖像去噪處理。通過實驗分析和定量計算與現(xiàn)有其他圖像去噪技術(shù)對比,結(jié)果表明,提出的GACA-MF改進(jìn)混合圖像去噪方法的效果優(yōu)于分別使用中值濾波算法和PCNN算法??梢?,利用自適應(yīng)的方式優(yōu)化網(wǎng)絡(luò)參數(shù)可以盡可能發(fā)掘PCNN的最大潛能。
關(guān)鍵詞:圖像去噪;遺傳蟻群算法;脈沖耦合神經(jīng)網(wǎng)絡(luò);中值濾波;優(yōu)化參數(shù)
中圖分類號:TP389 " " "文獻(xiàn)標(biāo)志碼:A " " " " "文章編號:2095-2945(2024)20-0001-07
Abstract: In order to realize adaptive image denoising, animproved pulse coupled neural network(PCNN) median filter image denoising method based on genetic ant colony algorithm (GACA) is proposed. Through the combination of genetic algorithm (GA) and ant colony optimization(ACO) algorithm, the calculation results of GA are used to enhance the early pheromones of ACO, and finally make ACO accelerate the optimization of the key parameters of PCNN in the positive feedback mechanism; then, PCNN was used to optimizemedian filtering technology for image denoising. Through experimental analysis and quantitative calculation, as well as comparison with other existing image denoising techniques, the results show that the proposed GACA-MF improved hybrid image denoising method is better than using median filtering algorithm and PCNN algorithm respectively. It can be seen that using adaptive way to optimize network parameters can explore the maximum potential of PCNN as much as possible.
Keywords: image denoising; genetic ant colony algorithm; pulse coupled neural network(PCNN); median filtering; Optimize parameters
近年來隨著數(shù)字技術(shù)的發(fā)展,數(shù)字圖像可以幫助人類更客觀更準(zhǔn)確地認(rèn)識世界。但是,數(shù)字圖像極易受椒鹽噪聲和隨機脈沖噪聲影響而使圖像質(zhì)量嚴(yán)重下降[1]。因此,如何在有效消除圖像噪聲的同時盡可能保留圖像細(xì)節(jié)信息和邊緣信息至關(guān)重要。
通常主要的噪聲類型有2種:椒鹽噪聲和隨機脈沖噪聲,本文主要針對含椒鹽噪聲的圖像進(jìn)行去噪研究。到目前為止,已經(jīng)提出了許多技術(shù)去除椒鹽噪聲。其中,中值濾波(Median Filter, MF)作為一種速度快、運算簡單的非線性濾波技術(shù),對椒鹽噪聲不僅具有良好的去噪效果,而且能很好地保護(hù)圖像邊緣信息,如文獻(xiàn)[2]提出一種基于小波變換的中值濾波圖像去噪方法,可在有效消除混合噪聲的同時保留圖像邊緣信息;……