黃有望 白新國



摘 要: 針對當(dāng)前警示標(biāo)志識別正確率低、識別速度慢等不足,提出基于視覺傳達(dá)的警示標(biāo)志識別方法。首先提取警示標(biāo)志識別的一些特征,然后結(jié)合人的視覺傳達(dá)感知設(shè)計一些規(guī)則,最后采用神經(jīng)網(wǎng)絡(luò)實現(xiàn)警示標(biāo)志識別,并通過具體實驗測試該方法的有效性和優(yōu)越性。測試結(jié)果表明,所提方法提高了警示標(biāo)志識別的正確率,有效改善了警示標(biāo)志識別的速度,而且警示標(biāo)志識別效果明顯優(yōu)于當(dāng)前其他警示標(biāo)志識別方法。
關(guān)鍵詞: 視覺傳達(dá); 警示標(biāo)志; 識別方法; 識別正確率; 識別速度; 神經(jīng)網(wǎng)絡(luò)
中圖分類號: TN911.73?34; TP181 文獻(xiàn)標(biāo)識碼: A 文章編號: 1004?373X(2018)11?0129?04
Warning signs recognition method based on visual communication
HUANG Youwang1, BAI Xinguo2
(1. College of Apparel & Design, Xian Polytechnic University, Xian 710048, China;
2. College of Computer Science, Xian Polytechnic University, Xian 710048, China)
Abstract: Aiming at the shortcomings of low recognition accuracy and slow recognition speed of warning signs, a warning signs identification method based on visual communication is proposed. The characteristics of warning signs recognition are extracted, and combined with the human′s visual perception to design some rules. And then the neural network is used to realize the warning signs identification. The effectiveness and superiority of the method were verified with specific test. The test results show that the method can improve the recognition accuracy and recognition speed of the warning signs, and its recognition effect is better than that of other warning signs recognition methods.
Keywords: visual communication; warning sign; identification method; recognition accuracy; recognition speed; neural network
隨著經(jīng)濟的迅速發(fā)展,城市化進(jìn)程速度不斷加快,城市道路和建筑越來越復(fù)雜,出現(xiàn)了許多警示標(biāo)記,這些警示標(biāo)記可以幫助人們迅速離開危險環(huán)境,以便更好地保護(hù)人身安全與財產(chǎn)安全。因此,對警示標(biāo)記進(jìn)行準(zhǔn)確識別具有十分重要的意義[1?3]。
針對警示標(biāo)記識別問題,學(xué)者們提出了許多有效的警示標(biāo)記方法,在警示標(biāo)記識別的研究中,首先要提取警示標(biāo)記識別特征,然后采用相應(yīng)的方法對警示標(biāo)記進(jìn)行分類和識別[4?6]。當(dāng)前警示標(biāo)記特征很多,如形狀特征、顏色特征等,單一特征一般無法獲得性能較優(yōu)的警示標(biāo)記識別結(jié)果,因此當(dāng)前主要采用多個特征實現(xiàn)警示標(biāo)記。警示標(biāo)記識別常用神經(jīng)網(wǎng)絡(luò)或者支持向量機進(jìn)行,其中支持向量機的學(xué)習(xí)速度比較慢,導(dǎo)致警示標(biāo)記識別的效率低,而神經(jīng)網(wǎng)絡(luò)的學(xué)習(xí)速度雖然快于支持向量機,但是識別效果有待進(jìn)一步提升[7?9]?!?br>