Gradient based adaptive thresholding

For images with poor and non-uniform illumination, adaptive thresholding is required to separate the objects of interest from the background. In this paper a new approach to create an adaptive threshold surface is proposed to segment an image. The technique is inspired by the Yanowitz’s method and is improved upon by the introduction of a simpler and more accurate threshold surface. The method is tested on several images of different patterns with varying illumination and the results are compared to the ones produced by a number of adaptive thresholding algorithms. In order to demonstrate the effectiveness, the proposed method had been implemented in medical and document images. The proposed method compares favorably against those using watershed and morphology in medical image and favorably against variable threshold and adaptive Otsu’s N-thresholding for document image.     

Acquisition of high-precision images for non-contact atomic force microscopy

The atomic force microscope (AFM) system has evolved into a useful tool for direct measurements of intermolecular forces with atomic-resolution characterization that can be employed in a broad spectrum of applications. The distance between cantilever tip and sample surface in non-contact AFM is a time-varying parameter even for a fixed sample height, and typically difficult to identify. A remedy to this problem is to directly identify the sample height in order to generate high-precision atomic-resolution images. For this, the microcantilever (which forms the basis for the operation of AFM) is modeled as a single mode approximation and the interaction between the sample and cantilever is derived from a van der Waals potential. Since in most practical applications only the microcantilever deflection is accessible, we will use merely this measurement to identify the sample height. In most non-contact AFMs, cantilevers with high-quality factors are employed essentially for acquiring high-resolution images. However, due to high-quality factor, the settling time is relatively large and the required time to achieve a periodic motion is long. As a result, identification methods based on amplitude and phase measurements cannot be efficiently utilized. The proposed method overcomes this shortfall by using a small fraction of the transient motion for parameter identification, so the scanning speed can be increased significantly. Furthermore, for acquiring atomic-scale images of atomically flat samples, the need for feedback loop to achieve setpoint amplitude is basically eliminated. On the other hand, for acquiring atomic-scale images of highly uneven samples, a simple PI controller is designed to track the desired constant sample height. Simulation results are provided to demonstrate the feasibility of the approach for both sample height identification and tracking the desired sample height.     

多波段图像融合算法研究及其新发展(Ⅰ)

对多通道、特别针对可见光与长波红外图像像素级融合的各种算法进行了深入的研究,针对工作实际需要,对其均进行了一定的改造和编程调试,对某些算法做了较重大改进,在若干方面提出了自己的具有特色的算法与思想。在此基础上编制了大型程序软件包。     

A fast and efficient color image enhancement method based on fuzzy-logic and histogram

A new fuzzy logic and histogram based algorithm for enhancing low contrast color images has been proposed here. The method is computationally fast compared to conventional and other advanced enhancement techniques. It is based on two important parameters M and K, where M is the average intensity value of the image, calculated from the histogram and K is the contrast intensification parameter. The given RGB image is converted into HSV color space to preserve the chromatic information contained in the original image. To enhance the image, only the V component is stretched under the control of the parameters M and K. The proposed method has been compared with conventional contrast enhancement techniques as well as with advanced algorithms. All the above techniques were based on the principle of transforming the skewed histogram of the original image into a uniform histogram. The performance of the different contrast enhancement algorithms are evaluated based on the visual quality, Tenengrad, CII and the computational time. The inter comparison of different techniques was carried out on different low contrast color images. Based on the performance analysis, we advocate that our proposed Fuzzy Logic method is well suited for contrast enhancement of low contrast color images.     

一种基于用户关注模型的立体图像舒适度客观评价方法

近年来,立体电视和立体电影的发展掀起了一股“3D”热潮。要让观众享受到高质量的立体内容,首先要解决人们在观看立体视频图像时的不舒适感。本文以立体图像的视差因素为切入点,结合人眼关注模型,设计了并实现了一种立体图像舒适度的客观评价方法。该方法根据人造立体图像的主观实验评价结果,建立了视差——舒适度模型,同时结合图像显著信息,对立体图像的舒适度进行了预测。对比主观评价结果,本文实现的所提出的立体图像舒适度客观评价方法所获取的评价结果与人眼感受具有较好的一致性。     

IR saliency detection via a GCF-SB visual attention framework

Infrared (IR) saliency detection with high detection accuracy is a challenging task due to the complex background and low contrast of IR images. In this paper, an IR saliency detection method via a new visual attention framework is proposed, which comprises two phases. In the first phase, a Gray & Contrast Features (GCF) model is established, in which the IR image is processed in two feature channels, a gray feature channel and a contrast feature channel. And then a primary feature map can be obtained by fusing the gray and contrast features from these two channels, which is the basis of the second phase. In the second phase, a Similarity-based Bayes (SB) model is established, in which two prior probabilities and two likelihood functions are calculated according to the previously obtained primary feature map. Finally, the saliency map is calculated with the obtained prior probabilities and likelihood functions by Bayes formula. Experimental results indicate that the proposed method can effectively reduce noise and enhance contrast of IR images with complex background and low contrast, and obtain a higher detection accuracy and robustness than seven state-of-the-art methods.     

基于可变形卷积神经网络的遥感影像密集区域车辆检测方法

车辆检测是遥感图像分析领域的热点研究内容之一,车辆目标的智能提取和识别,对于交通管理、城市建设有重要意义。在遥感领域中,现有基于卷积神经网络的车辆检测方法存在实现过程复杂并且对于车辆密集区域检测效果不理想的缺陷。针对上述问题,该文提出基于端到端的神经网络模型DF-RCNN以提高车辆密集区域的检测精度。首先,在特征提取阶段,DF-RCNN模型将深浅层特征图的分辨率统一并融合;其次,DF-RCNN模型结合可变形卷积和可变形感兴趣区池化模块,通过加入少量的参数和计算量以学习目标的几何形变。实验结果表明,该文提出的模型针对密集区域的车辆目标具有较好的检测性能。     

基于改进多尺度间隙度的海上目标分割方法

针对复杂海面背景下的可见光图像舰船目标分割问题,引入了图像的多尺度间隙度特征,分析了Dong提出的图像间隙度特征提取算法存在的问题,并进行了改进,采用滑动盒子内像素值方差计算滑动盒子的质量,提高了舰船目标与海面背景的可分离程度,最后利用改进算法提取图像的多尺度间隙度特征对海上舰船目标进行分割。实验结果表明,利用改进方法对海面背景下舰船目标进行分割,相对于sobel边缘检测、ostu阈值分割和基于传统提取算法的多尺度间隙度特征分割方法结果更优,可以更好地抑制海面亮度变化、海浪和背景杂波的影响,准确分割出海上舰船目标。     

基于独立成分分析的多时相遥感图像变化检测

变化检测是通过分析多时相遥感图像间的差异实现地物变化信息的提取,而消除多时相遥感图像中的相关性是提取变化信息的一种有效途径。独立成分分析(ICA)作为近年出现的盲源分离技术,能够有效地消除多源信号间的二阶和高阶相关,经其变换的各分量之间相互独立。该文提出一种应用ICA变换实现多时相遥感图像变化检测的算法,首先对多时相多光谱遥感图像进行独立成分分析,得到彼此没有相关信息的独立成分,并且各独立成分图像中的变化信息得到增强;然后通过分析变换后的独立成分实现地物的变化检测。实验结果显示该算法比传统的方法具有更好的性能。     

基于八邻域判决的红外运动小目标检测方法

针对空中红外图像中运动小目标的特征,提出了一种基于八邻域判决的运动小目标检测方法。该方法根据小目标运动的连续性,对前后两帧图像中小目标的位置进行比较,依据八邻域判决算法确定运动小目标的运动方向,从而将小目标与高频噪声点区别出来。实践证明,该技术结构简单,运算速度快,是一种检测性能好且在工程上易于实现的方法。