This is the source code to SEEDS superpixelsThis is the source code to SEEDS superpixels, as described in M. Van den Bergh, X. Boix, G. Roig, B. de Capitani and L. Van Gool, "SEEDS: Superpixels Extracted via Energy-Driven Sampling", ECCV 2012.
The SEEDS class is defined in seeds.h and seeds2.cpp. The method can be run on an RGB image (stored in an UINT* array) by using the following constructor, initialization and iteration methods:
SEEDS(int width, int height, int nr_channels, int nr_bins, int min_size);
void initialize(UINT* image, int seeds_w, int seeds_h, int nr_levels);
void iterate();
The initial size of the superpixels is independent from the image size, and defined by the following 3 parameters: , and . The first two define the minimum block size in the SEEDS algorithm. The defines the number of times this block size is doubled to
自编的BP神经网络Matlab程序,可完美运行。还附有0到9训练样本集,供大家参考学习(Self BP neural network Matlab program can be run perfectly. Accompanied by 0-9 training sample set for your learning)
用MATLAB软件产生一个被高斯噪声污染的lenna图像,使用T3、T5、高斯滤波、中值滤波进行处理;(MATLAB software with a Gaussian noise pollution Lenna image, the use of T3, T5, Gaussian filtering, median filtering for processing )
频域波束成形算法,适用于所有一维和二维阵列的波束成形计算。对于采样时间较短的信号,采用了2次分段,对于采样时间足够的信号则只需要1次分段计算。计算结果包含成像图结果和各扫描点处的频谱结果(Frequency domain beamforming algorithm, applicable to all one-dimensional and two-dimensional array beam forming calculations. For shorter sampling signal, using a second segment, the sampling time sufficient signal need only 1 segment calculations. Calculation results include the results of spectral imaging of the results of each scan point)