The multiple subimage were extracted from the reference image according to a predefined selection method. They were regarded as a template and correlated with the real image. The multiple matching results were integra...
详细信息
The multiple subimage were extracted from the reference image according to a predefined selection method. They were regarded as a template and correlated with the real image. The multiple matching results were integrated with the clustering method based on the spatial relation among the subimages. The confidence of the final matching result was given. The test result showed that the method had improved the performance of correlation-based matching.
The common method for image matching is correlation matching. The calculating number was very large when correlation function was used between two matching images. Two steps approach was proposed for image matching th...
详细信息
The common method for image matching is correlation matching. The calculating number was very large when correlation function was used between two matching images. Two steps approach was proposed for image matching that was coarse and exact way. Firstly every pixel's gray level was summed up along every column to calculate the mean of column pixels. Then one dimension signal was coarsely matched using one order correlation function. Secondly the points with larger correlation matching values were chosen as the better matching points. These less points were used in the original image as the coordinate origins to calculate the two orders correlation functions. The point with maximum correlation value would be the optimal matching position. Experiences with three kinds of images prove that this approach is faster than conventional one and the matching result is accurate.
Moment invariants are important shape descriptors in computer vision. The method of generating trigonometric function is suggested as a new efficient way to derive various moment invariants. General rule of moment inv...
详细信息
Moment invariants are important shape descriptors in computer vision. The method of generating trigonometric function is suggested as a new efficient way to derive various moment invariants. General rule of moment invariant construction is summarized and the notion of moment invariant space is established. Five more moment invariants are derived and tested. Furthermore, general explicit constructions of some high-order moment invariants are developed. The comparisons are made between the stabilities of several invariants for the discrete image. With the help of these additional moment invariants, we can distinguish the image and the object more accurately.
Aiming at the nonlinearity of photoresponse characteristic in infrared focal plane array (IRFPA), an approach for nonuniformity correction in IRFPA was proposed, which is simple and easy to implement by hardware circu...
详细信息
Aiming at the nonlinearity of photoresponse characteristic in infrared focal plane array (IRFPA), an approach for nonuniformity correction in IRFPA was proposed, which is simple and easy to implement by hardware circuit. The theoretical analysis and comparison of simulation results show that its performance is more perfect than that of the piecewise linear algorithm and the polynomial fitting algorithm, and only a few correction factors are needed.
This paper proposed a method for generating chaotic key stream based on the chaotic map and a fast algorithm to implement it. This method divides the chaotic attractor into 2n symmetric sub-intervals, samples from the...
详细信息
This paper proposed a method for generating chaotic key stream based on the chaotic map and a fast algorithm to implement it. This method divides the chaotic attractor into 2n symmetric sub-intervals, samples from the chaotic orbit every n time intervals, and gets the independently and uniformly distributed 2n-phase key stream. The theoretical and numerical analyses show that the sequence also has a high complexity. This method can be used in fields such as cryptography, numerical simulation and spread spectrum communication, etc.
This paper reviews the AIM 2020 challenge on extreme image inpainting. This report focuses on proposed solutions and results for two different tracks on extreme image inpainting: classical image inpainting and semanti...
详细信息
This paper proposes a genetic-based algorithm for surface reconstruction of three-dimension (3-D) objects from a group of contours representing its section plane lines. The algorithm can optimize the triangulation of ...
详细信息
This paper proposes a genetic-based algorithm for surface reconstruction of three-dimension (3-D) objects from a group of contours representing its section plane lines. The algorithm can optimize the triangulation of the surface of 3-D objects with a multi-objective optimization function to meet the needs of a wide range of applications. Further, a new crossover operator for triangulation and a new 3-D quadrilateral mutation operator are also introduced.
To solve the heterogeneous image scene matching problem, a non-linear pre-processing method for the original images before intensity-based correlation is proposed. The result shows that the proper matching probability...
详细信息
To solve the heterogeneous image scene matching problem, a non-linear pre-processing method for the original images before intensity-based correlation is proposed. The result shows that the proper matching probability is raised greatly. Especially for the low S/N image pairs, the effect is more remarkable.
A FLIR image segmentation algorithm based on genetic algorithm and fuzzy set theory was presented. The method defines different member function for the object and background of the image to transform the image into fu...
详细信息
A FLIR image segmentation algorithm based on genetic algorithm and fuzzy set theory was presented. The method defines different member function for the object and background of the image to transform the image into fuzzy domain with maximum fuzzy entropy. The procedure for finding combination of a, b and c is implemented by genetic algorithm, thresholding image into object and background by maximizing the fuzzy entropy. The experiment results show that our proposed method gives better performance and higher calculation speed than other general methods with good real-time by using genetic algorithm.
Contrastive learning, which aims to capture general representation from unlab.led images to initialize the medical analysis models, has been proven effective in alleviating the high demand for expensive annotations. C...
详细信息
暂无评论