作业帮 > 英语 > 作业

英语翻译帮忙把下段文字翻译成英文 目标退化的原因分析:在归一化的神经网络算法中,再求取像元(i,j)的期望校正输出时,直

来源:学生作业帮 编辑:作业帮 分类:英语作业 时间:2024/07/13 17:38:05
英语翻译
帮忙把下段文字翻译成英文
目标退化的原因分析:在归一化的神经网络算法中,再求取像元(i,j)的期望校正输出时,直接采用上次校正输出的(i,j)四邻域均值.对于场景中同一区域内部由于信号在空间上是连续渐变的,这样的期望输出是合理的.对于紧靠边缘的像素点,通过这种期望值的算法就会在一定程度上弱化图像不同区域间的边缘对比度.当图像在运动过程中是,图像边缘不会固定在某一像素点上,因此,这种弱化不会持续下去,相对于较长的校正帧,这种弱化可以忽略.但是当场景图像静止时,图像边缘就会在某一像素固定下来,边缘的弱化就会持续下去.不同区域内的能量就会通过邻域均值互相扩散,造成静止目标的退化.
Marked degradation analysis:normalized at the neural network algorithm,and then strike a pixel (i,j) Calibration output expectations,the direct use of Last Calibration output (i,j) around the domain mean.For scenes in the same region in space because of the signal on a continuous gradient,so that the desired output is reasonable.For close to the edge of the pixel points,through which expectations of the algorithm will be weakened to some extent in different regions of the image contrast between the edge.When the images are in the process of exercise,image edge pixel will not be fixed at a certain point on,so this will not continue to weaken,as opposed to longer Calibration frame,this weakening can be ignored.But when scenes of static images,the image edge will be fixed at a certain pixel,the edge will continue to weaken.In different regions of energy would mean each other through the proliferation of neighborhood,resulting in the degradation of stationary targets.