融合边缘检测与区域生长的交通图像分割方法

Traffic Image Segmentation Method Integrating Edge Detection and Region Growing

刘浩[1] 董超俊[1,2]

在交通监控中,如何从复杂的背景中分割运动物体是至关重要的一步,针对车辆的运动阴影对图像分割产生的不利影响,提出了一种新的融合边缘检测与区域生长的彩色图像分割算法,算法同时考虑了图像的彩色信息和空间信息。该算法首先对彩色图像边缘检测,并根据检测结果设置种子像素;再基于颜色相似性生长准则,结合边缘检测结果,对每个种子点进行区域生长;最后,利用区域合并算法对剩余的像素进行合并。实验结果表明该算法很大程度上克服了阴影给图像分割带来的不利影响。[著者文摘]

Segmentation of moving object image with complex background is a crucial step in traffic surveillance. In order to solve the problem of the negative effects of moving object shade on a color image segmentation, a novel algorithm for color image segmentation integrated edge detection and region growing is proposed. Both the color information and spatial information of traffic color image are taken into account in this algorithm. Firstly, the edge of color image is detected by using a simple edge detection algorithm, and then, the seed pixels are selected in terms of the detection results of image. Secondly, seed pixel region growing is performed based on the hue similarity criterion and the results of edge detection. Finally, region merged algorithm is used to solve the remaining pixels that have not been classified into regions. Experimental results show that the negative effects of shading in a traffic color image on segmentation can be overcome to a great extent by using the proposed algorithm.[著者文摘]

基于均值偏移的灰度图像分割方法

Gray Image Segmentation Algorithm Based on Mean Shift

李正周 彭素静 王允 刘国金

针对灰度图像分割往往存在的过分割或欠分割问题，提出了一种基于均值偏移的图像分割方法．该方法通过联合像素的空间位置和灰度特征，建立图像的特征矢量，构造基于像素的位置和灰度的改进核函数直方图．采用均值偏移算法搜索图像的特征模式，并以此对图像进行平滑滤波和分割．对以地物为背景的图像分割结果表明，该方法既能抑制具有纹理的大片背景，又能提取出面积较小且轮廓清晰的物体，分割的物体完整且符合人眼视觉，较有效地避免了图像过分割和欠分割的问题．[著者文摘]

The new image segmentaiotion algorithm based on mean shift is proposed to overcome such problems as over-segmentation or lack-segmentation, and to improve the performance of gray image segmentation. It constructs a novel kernel function histogram by combing the position and the gray value of the pixel, and then makes use of mean shift with this new kernel function to automatically search the models of the features, and to segment the gray image. Experiments based on the ground as the main background image segmentation are carried out by the Canny,Sobel and this mean shift,and the results show that the proposed algorithm not only suppress the big object and small object effectively,and it could effectively avoids image over-segmentation or lack-segmentation.[著者文摘]

一种C-均值聚类图像分割的模糊熵后处理方法

Fuzzy Entropy Based Post-Processing Method for C-Mean Clustering Image Segmentation Algorithm

赵凤 范九伦

提出了一种结合C-均值聚类算法和模糊熵的图像分割方法，该方法先采用C均值聚类算法对含噪图像进行初步分割，再利用模糊熵准则作后续处理。该方法一方面能够继承C-均值聚类算法的优点，可以灵活地用在基于多特征和多阂值的图像分割中，另一方面充分考虑了图像的区域信息，利用模糊熵最小作为准则，对c均值聚类算法初步分割结果的错分类点作了进一步的处理，克服了C-均值聚类算法对噪声敏感的缺点。实验结果表明，本文方法在运算开销上只比C-均值聚类算法多4～6S，对于低信噪比的图像能够取得优于C-均值聚类算法的分割效果。[著者文摘]

An image segmentation method for combining C mean clustering algorithms and fuzzy entropy is presented. Firstly, the pre-segmentation on the image is made by one of the C- mean elustering algorithms; then further proeessing is clone by using fuzzy entropy principle. The method inherits the advantages of the C-mean elustering algorithms, that is, it ean be eas fly applied to image segmentation tasks with multi-feature and multi-threshold. Furthermore, the method considers the region information of the image, and utilizes the minimum fuzzy entropy prineiple to post-proeess the wrong classified points of the pre-segmentation result. Thus the method ean overeome the disadvantage of the C-mean elustering algorithms, that is, it is not sensitive to the noise. Experimental results show that the CPU-time of the method is only 4- 6 s more than the C-mean elustering algorithms, and it ean behave better in segmenting images of low signal to noise ratio than the C mean clustering algorithms.[著者文摘]

月面巡视探测器的图像分割及识别方法

Image segmentation and recognition of lunar rover

石德乐[1] 叶培建[2] 贾阳[2] 王荣本[3] 郭烈[3]

针对月面巡视探测器的识别技术，通过彩色图像识别的办法解决了在月面环境存在较强阴影区域对月面巡视探测器识别带来的影响，进行了多种彩色图像分割方法的比较，提出了多通道彩色分量融合的图像分割方法，并用线性分类器的模式识别理论对其原理进行了分析，用形态学滤波方法对分割后的图像进行了滤波，并用分割区域标识算法进行了识别区：战的标识。试验表明，该方法能够适应各种光照条件的影响，具有较强的鲁棒性，对各种噪声有明显的抑制作用。该方法使彩色图像分割技术变得简单、可靠，并具有一定的通用性，可以应用到未来的月球探测中去。[著者文摘]

In lunar exploration, the lunar rover should be recognized by the lunar lander, and the technique of lunar rover recognition from the lunar lander was studied. In order to eliminate the effect of the strong shadow in the lunar illumination condition on the rover recognition, the color segmentation method was used to perform the grey image processing. Based on the comparison of several color image segmentation algorithms, a technique of multi-channel color threshold fusion was proposed, and its principle was analyzed by the mode recognition theory of linear classifier. The segmented image was filtered by the morphologic operator and the recognized region was labelled by the label algorithm. The experimental results show that the proposed technique is robust and compatible with the varied illumination, can restrain the background noises, may be used in future lunar exploration. It makes the color segmentation simple and reliable and can also be generalized to common color segmentation application.[著者文摘]

改进加权FCM用于裂纹蛋2D图像分割的研究

Improved Weighted FCM in Crack Egg 2D Image Segmentation

张莉 郁志宏

解决裂纹鸡蛋图像灰度直方图目标与背景区域分布模糊、图像分割效果差的问题。通过将包含空间信息的二维直方图和改进特征加权FCM算法有机结合，迭代寻求最佳聚类有效性函数和加权矩阵，实现鸡蛋图像缺陷分割。同时，对经典FCM和改进特征加权FCM算法的性能进行了分析比较。结果表明：提出的算法更接近于真实聚类中心，目标函数值亦得到改善；二维直方图的改进特征加权模糊聚类算法更好地提取了裂纹鸡蛋图像的细节信息，图像分割效果好。[著者文摘]

Aim to improve image segmentation effect on crack inspection of eggs. Taking the 2D histogram integrated with spatial information as the sample of improved FCM algorithm with weighted features, iteratively seeking optimal validity function and weighted matrix, crack egg segmentation is achieved. Comparing with the performance of classical FCM, the results show that the algorithm of the improved FCM with weighted features has a better object function and an adjacent clustering data point. The algorithm of improved FCM with weighted features performed in 2D histogram could catch more detail information of image, and the segmentation effect on crack egg image is good.[著者文摘]

基于MATLAB的茄子图像分割方法

Image Segmentation for Eggplant Based on MATLAB

姚立健 丁为民 刘璎瑛

针对茄子图像的灰度和颜色特点,利用MATLAB中丰富的图像处理函数,分别进行了色差分割和色调分割。在色调分割中,采用了自动选取阈值的Otsu法。在去除残留噪音的处理中,采用标注的方法对二值图像的各连通区域进行面积统计。保留最大面积的区域,从而使分割效果大大改善。利用多参数来衡量分割效果,使评价做到最大程度的客观、合理。[著者文摘]

Based on the gray - level and color characteristics of eggplant image, hue and color - difference segmentation were conducted by using the rich image processing functions embedded in MATLAB. The auto threshold - adaptive method of Otsu operation was specially used for hue segmentation. And upon the elimination of the residue noise, labeling method for statistical calculation was introduced for the connected regions of the binary image. The maximum areas were preserved to improve segmentation effects. In addition, multi -indices were applied to assess the effect of segmentation, in order to get impersonal and reasonable assessment.[著者文摘]