小户型室内空间视觉合理性评价仿真  

Evaluation of the Visual Rationality of Small Apartment Interior Space

作　　者：胡家强 HU Jia-qing(College of Optical and Electronical Information,Changchun University of Science and Technology,Changchun 川in 130000,China)

机构地区：[1]长春理工大学光电信息学院

出　　处：《计算机仿真》2019年第7期339-342,共4页Computer Simulation

摘　　要：针对当前室内空间视觉合理性评价方法存在评价精度低、复杂度高的问题,提出基于最小二乘法的小户型室内空间视觉合理性评价方法。使用针孔相机采集室内空间图像,并建立图像集。设置初始图像分割阈值和图像中的扫描点,以此计算图像在横纵方向上的一阶差分。利用一阶差分得到室内空间图像梯度值和梯度模,根据图像梯度模得到边缘像素梯度均值。使用图像初始分割阈值和图像边缘像素的梯度均值计算图像自适应分割阈值,实现室内空间图像的最终分割。对分割结果中离散数据点之间的距离进行计算,得到图像中的连续特征点。根据连续特征点集合获取图像中存在线性关系的线段点集,通过最小二乘法获取室内空间图像坐标原点至线段的距离、图像坐标原点至线段垂直于横轴的夹角最优值。将经过最优值的线段定义为可最大程度表征图像中各点关系的线段,利用比较其与图像中存在线性关系的点集间关系,判断室内空间视觉合理性。实验表明,上述方法评价精度高,运行复杂系数平均为0.2。Currently, the evaluation method for visual rationality of indoor space has some problems of low evaluation accuracy and high complexity. Therefore, this paper focuses on a method to evaluate the visual rationality of indoor space in small apartment based on least squares method. At first, the pinhole camera was used to collect indoor space images and create the set of images. Secondly, the initial image segmentation threshold and the scanning point in image were set to calculate the first-order difference of image in the horizontal and longitudinal directions. Then, the first-order difference was used to obtain the gradient value and gradient modulus of indoor space image. According to the image gradient modulus, the gradient mean value of edge pixel was obtained. Moreover, the initial segmentation threshold of image and the gradient mean value of edge pixel were used to calculate the adaptive segmentation threshold of image, so as to achieve the final segmentation of indoor space image. Meanwhile, the distance between the discrete data points in the segmentation result was calculated to obtain continuous feature point in image. According to the set of continuous feature points, the set of line segment point with linear relation in the image was obtained. In addition, the least squares method was used to obtain the distance from the coordinate origin of interior space image to the line segment and the optimal value of angle between the image coordinate origin and the line segment perpendicular to the horizontal axis. Finally, the line segment with the optimal value was defined as the line segment which could characterize the relationship of each point in image as much as possible. By comparing with the relationship between the sets of points with linear relationship in image, the visual rationality of indoor space could be judged. Simulations show that the proposed method has high evaluation accuracy. The average running complexity coefficient is 0. 2.

关 键 词：室内空间 视觉合理性 评价 最小二乘 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]