Then the research of image enhancement based on of plateau equalization algorithm and genetic algorithm is done then in this chapter . image segmentation with algorithm of 2d - maximal variance between - class and region growing is accepted to eliminate the background and the isolated noise points which can not be the target 論文在紅外目標(biāo)檢測方面,以對經(jīng)過預(yù)處理的圖像進行分割為切入點,采用二維類間方差法和局域生長法來進行圖像的分割,消除了原始圖像中一些不可能為目標(biāo)的孤立噪聲點和大面積的背景區(qū)域。
To sum up , the main research work and its characteristic in this dissertation includes as follow : firstly , introduce the research background , point out the aim and actual meaning of the research , summarize the research status and development trend of time domain equalization algorithms and system on a chip design international and oversea 概括起來,本文的主要研究工作及其特點包括: 1首先介紹了課題的研究背景,指出了本文選題的目的和實際意義,總結(jié)了時域均衡算法和片上系統(tǒng)設(shè)計的國內(nèi)外研究現(xiàn)狀和發(fā)展趨勢。
This paper formulize the foundational principle and development of blind equalization technique based on neural network . the traditional neural network blind equalization algorithms have many advantages and disadvantages . through analyse these disadvantage , two kinds of new blind equalization algorithm based on neural network is proposed 本文首先對基于神經(jīng)網(wǎng)絡(luò)的盲均衡技術(shù)原理及發(fā)展進行了闡述,分析了原有神經(jīng)網(wǎng)絡(luò)盲均衡算法的優(yōu)缺點,然后對原有算法的缺陷進行改進,提出了兩類新的神經(jīng)網(wǎng)絡(luò)盲均衡算法。
The second is blind equalization algorithm based on bili near recurrent neural network . this algorithm adopt the bilinear recurrent neural network and design a new transmission function and cost function . through computer simulation , all proposed algorithm have better convergence performance 第二類是基于反饋神經(jīng)網(wǎng)絡(luò)( rnn )的盲均衡算法,此類算法使用了一種新型的雙線性反饋神經(jīng)網(wǎng)絡(luò)( blrnn ) ,并將這種網(wǎng)絡(luò)結(jié)構(gòu)擴展到了復(fù)數(shù)域內(nèi),然后依據(jù)盲均衡的特點為網(wǎng)絡(luò)設(shè)計了傳輸函數(shù)和代價函數(shù)。
In this proposed algorithm , the phase rotation can be compensated using higher order statistics , nonlinear transform of equalizer outputs , and decision feedback algorithm , the faster convergent speed can be obtained by introducing sign algorithm , and the mse can be reduced via employing decision feedback equalization algorithm 該算法充分利用高階統(tǒng)計量所包含的相位信息、均衡器輸出信息的非線性變換及判決反饋算法來補償相位旋轉(zhuǎn);利用符號算法可以減少計算量的特點來加快收斂速度;利用判決反饋濾波器的性能來減小均衡器輸出的均方誤差。
In this case , time domain equalization algorithm based on channel estimation will influence equalization effect if choosing noise sensitive algorithm to estimate channel , so it is necessary to adopt an anti - noise and high estimation precision estimation algorithm . in this dissertation , a new time domain equalization algorithm based on low rank hankel approximati Dsl線路實際工作環(huán)境存在各種干擾,尤其當(dāng)多用戶時,串音干擾很嚴重,對于采用基于信道估計的時域均衡算法,如果使用對噪聲敏感的算法進行信道估計,必然影響均衡效果,因此需要采用具有抗噪聲,估計精度高