Due to the ambiguity of text and the lack of location information in training data， current state-of-the-art diffusion model cannot accurately control the locations of generated objects in the image under the condition of text prompts. To address this issue， a spatial condition of the object’s location range was introduced， and an attention-guided method was proposed based on the strong correlation between the cross-attention map in U-Net and the image spatial layout to control the generation of the attention map， thus controlling the locations of the generated objects. Specifically， based on the Stable Diffusion （SD） model， in the early stage of the generation of the cross-attention map in the U-Net layer， a loss was introduced to stimulate high attention values in the corresponding location range， and reduce the average attention value outside the range. The noise vector in the latent space was optimized step by step in each denoising step to control the generation of the attention map. Experimental results show that the proposed method can significantly control the locations of one or more objects in the generated image， and when generating multiple objects， it can reduce the phenomenon of object omission， redundant object generation， and object fusion.

In order to support the distributed transmission of a lot of tasks on the data exchange platform for civil aviation information, it needs to establish the efficient task scheduling algorithms and models. Based on the infrastructure and needs of the platform, after analyzing the existing task scheduling models and scheduling algorithms, a new task scheduling model was proposed to fulfill the data exchange on this platform. This model mapped the point-to-multipoint data transmission network to a Steiner tree problem with delay and bandwidth constraints, and an improved Genetic Algorithm (GA) was also proposed to solve the constrained Steiner tree problem. The results of comparative experiment with the maximum bandwidth allocation algorithm prove the validity and feasibility of the proposed model.

Traditional image object detection algorithm based on local feature is sensitive to rotation and occlusion; meanwhile, it also obtains low detection precision and speed in many cases. In order to improve the performance of this algorithm, a new image objects detection method applying objects’ local feature to sparse representation theory was introduced. Employing supervised random tree method to learn local features of sample images, a dictionary could be formed. The combination of sub-image blocks of test image and well trained dictionary in first stage could predict the location of the object in the test image, in this way it could obtain a sparse representation of the test image as well as the object detection goal. The experimental results demonstrate that the proposed method achieves robust detection results in rotation, occlusion condition and intricate background. What’s more, the method obtains higher detection precision and speed.

Current survivability models can hardly bring up a practical solution, nor reflect the properties of Web applications. Firstly, the properties of Web applications were analyzed, especially the differences between atomic Web applications and composite Web applications. Secondly, the mathematical model reflecting the invoking relationship of the atomic Web applications for the development of composite Web application was constructed. Lastly, a survivability model for atomic Web applications and a Markov-based survivability model for composite Web applications with regard to runtime environment were proposed. And on the basis of these models, a recovery approach for Web applications was given, when part or all of the functions failed in adverse environment. Besides, a case was analyzed using these models, and its recovery procedures were given, in which the high survivability was guaranteed.

Shortcomings of image encryption schemes based on chaos currently can be mainly classified into three types as follows: firstly, low dimensional chaotic sequences cause degradation of chaos; secondly, the structure of chaotic system adopted is too simple; thirdly, the algorithm is merely related to the structure of chaotic system and key. Concerning the shortcomings above, this paper presented a new image group encryption algorithm. In order to overcome the degradation of chaos, the algorithm generated pixel diffusion matrix by coupling chaotic sequences of high dimensional hyperchaos system with chaotic sequences of one dimensional chaotic mapping by means of matrix tensor product. In the process, the generation of pixel diffusion matrix was controlled by plaintext, thus the algorithm was related to plaintext, which enhanced the security of the algorithm.

The identifying, analyzing and evaluating algorithm of finding distance-based Outlier (Cell-Based) was firstly studied, and its advantages and disadvantages were pointed out. And then, a new Outlier finding algorithm-algorithm of finding Outlier for reclustering based on distance was proposed. Theoretical analysis and experimental results show that this algorithm can not only effectively overcome the faults of traditional Cell-Based algorithm, i.e. need to be recomputed from scratch for every change of the parameters, and only suitable for finding the Outlier of low dimension, but also obviously avoid the problems caused by randomly selecting initial value to produce different finding results of Outlier at higher convergence speed.