科技通报
科技通報
과기통보
BULLETIN OF SCIENCE AND TECHNOLOGY
2015年
8期
251-253,276
,共4页
色彩调节%舒适度%适应模型
色綵調節%舒適度%適應模型
색채조절%서괄도%괄응모형
color adjustment%comfort%adaptation model
在建立夜间建筑景观色彩调节的舒适度适应模型过程中,容易忽略人眼立体视觉注意力机制,导致传统的舒适度适应模型由于不符合人眼的立体视觉注意力机制,无法有效建立夜间建筑景观色彩调节的舒适度适应模型,提出一种基于立体视觉技术的夜间建筑景观色彩调节的舒适度适应模型,分析了静态视觉显著特征和夜间建筑景观立体图像的视差,求出静态显著图和深度显著图,将其结合成立体显著图,为夜间建筑景观色彩调节舒适度适应模型提供可靠依据.在立体显著图的基础上对夜间建筑景观色彩特征矢量进行采集,随机选取夜间建筑景观图像集合中若干对立体图像组成训练样本数据集合,通过支持向量回归完成训练样本数据集合的训练,获取最佳权重矢量以及最佳偏置量,得到最佳支持向量回归训练模型,从而实现夜间建筑景观立体图像舒适度适应模型的建立.仿真实验结果表明,所提模型具有很高的准确性.
在建立夜間建築景觀色綵調節的舒適度適應模型過程中,容易忽略人眼立體視覺註意力機製,導緻傳統的舒適度適應模型由于不符閤人眼的立體視覺註意力機製,無法有效建立夜間建築景觀色綵調節的舒適度適應模型,提齣一種基于立體視覺技術的夜間建築景觀色綵調節的舒適度適應模型,分析瞭靜態視覺顯著特徵和夜間建築景觀立體圖像的視差,求齣靜態顯著圖和深度顯著圖,將其結閤成立體顯著圖,為夜間建築景觀色綵調節舒適度適應模型提供可靠依據.在立體顯著圖的基礎上對夜間建築景觀色綵特徵矢量進行採集,隨機選取夜間建築景觀圖像集閤中若榦對立體圖像組成訓練樣本數據集閤,通過支持嚮量迴歸完成訓練樣本數據集閤的訓練,穫取最佳權重矢量以及最佳偏置量,得到最佳支持嚮量迴歸訓練模型,從而實現夜間建築景觀立體圖像舒適度適應模型的建立.倣真實驗結果錶明,所提模型具有很高的準確性.
재건립야간건축경관색채조절적서괄도괄응모형과정중,용역홀략인안입체시각주의력궤제,도치전통적서괄도괄응모형유우불부합인안적입체시각주의력궤제,무법유효건립야간건축경관색채조절적서괄도괄응모형,제출일충기우입체시각기술적야간건축경관색채조절적서괄도괄응모형,분석료정태시각현저특정화야간건축경관입체도상적시차,구출정태현저도화심도현저도,장기결합성입체현저도,위야간건축경관색채조절서괄도괄응모형제공가고의거.재입체현저도적기출상대야간건축경관색채특정시량진행채집,수궤선취야간건축경관도상집합중약간대입체도상조성훈련양본수거집합,통과지지향량회귀완성훈련양본수거집합적훈련,획취최가권중시량이급최가편치량,득도최가지지향량회귀훈련모형,종이실현야간건축경관입체도상서괄도괄응모형적건립.방진실험결과표명,소제모형구유흔고적준학성.
On the night landscape architecture color adjust the comfort of adaptation in the process of model is set up, easy to overlook the human eye stereo visual attention mechanism, lead to the traditional comfort to adapt to the model, because do not conform to the human eye stereo visual attention mechanism, unable to effectively build night landscape architecture color adjustment of comfort to adapt to the model, put forward a kind of night landscape architecture color correction based on stereo vision comfort to adapt to the model, analyzes the static visual characteristic and night landscape architecture stereo parallax images, and the static significant figure and depth of significant figure, will be combined into the three-dimensional figure, in order to adapt night landscape architecture color adjustment comfort model to provide reliable basis.On the basis of the three-dimensional figure significantly on the characteristic vector of night landscape architecture color, architectural landscape image collection of several random night of three-dimensional image of the training sample data collection, through the training of the support vector regression to complete the training sample data set, the optimal weight vector and the optimal offset, get the best training support vector regression model, so as to realize the night landscape architecture comfort to adapt to the model of three-dimensional images.The simulation results show that the proposed model has a high accuracy.