图片旋转对于识别模式带来的变化

Posted 2022-02-06 卓晴

tags:

篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了图片旋转对于识别模式带来的变化相关的知识，希望对你有一定的参考价值。

简介：测试了七段数字图片段角度对于识别的影响。可以看到对于6,9两个数字，在旋转过程中他们之间会相互转换。2,5在旋转180°之后，它们与自己相同，会出现两个识别相同。0,8与其他数字自检有着比较多的差异。 1,7两个数字有着较强的交叉。

关键词： 七段数字，LCD

▲ 图1 旋转的模式

§01 旋转图片

在博文 模型扫描识别图片 对于 七段数码数字模型 进行扫描测试，也就是利用对七段数码图片进行扫描识别，给出了波动的结果。

下面测试一下数字旋转对应的输出结果。

1.1 旋转图片

选择下面LED图片中的数码图像作为旋转测试。

▲ 图1.1.1 用于旋转测试的图片

1.1.1 分割形成旋转图片

（1）分割代码

from headm import *                 # =
import cv2
from paddle.vision.transforms import rotate

backgroundfile = '/home/aistudio/work/7seg/rotate/220112235650.BMP'
img = cv2.imread(backgroundfile)
printt(img.shape)

backcolor = mean(mean(mean(img, axis=0), axis=0))
printt(backcolor)

imgfile1 = '/home/aistudio/work/7seg/rotate/052-01234.JPG'
imgfile2 = '/home/aistudio/work/7seg/rotate/053-56789.JPG'

cutdir = '/home/aistudio/work/7seg/rotate/cutdir'

outsize = 48
rotateangle = range(0, 360, 10)

def cutimg(imgfile):
    img = cv2.imread(imgfile)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

    imgwidth = gray.shape[1]
    imgheight = gray.shape[0]

    for i in range(5):
        left = int(imgwidth * i // 5)
        right = int(imgwidth * (i + 1) // 5)
        data = gray[:, left:right]
        data = cv2.resize(data, (outsize,outsize))

        for r in rotateangle:
            bgdim = ones((512,512)) * backcolor
            n1 = 256-24
            n2 = 256+24
            bgdim[n1:n2, n1:n2] = data

            rotdata = rotate(bgdim, r)
            data1 = rotdata[n1:n2,n1:n2]

            fn = os.path.join(cutdir, '%d_%03d.BMP'%(i+5, r))
            cv2.imwrite(fn, data1)

cutimg(imgfile2)

（2）分割旋转结果

▲ 图1.1.2 旋转后的结果

from headm import *                 # =
import cv2

cutdir = '/home/aistudio/work/7seg/rotate/cutdir'
filedim = os.listdir(cutdir)

rotateangle = range(0, 360, 10)

gifpath = '/home/aistudio/GIF'
if not os.path.isdir(gifpath):
    os.makedirs(gifpath)
gifdim = os.listdir(gifpath)
for f in gifdim:
    fn = os.path.join(gifpath, f)
    if os.path.isfile(fn):
        os.remove(fn)

count = 0
for r in rotateangle:
    plt.clf()
    plt.figure(figsize=(10, 4))
    for i in range(2):
        for j in range(5):
            n = i*5+j
            fn = os.path.join(cutdir, '%d_%03d.BMP'%(n, r))

            img = cv2.imread(fn)
            plt.subplot(2,5,i*5+j+1)
            plt.imshow(img)

    savefile = os.path.join(gifpath, '%03d.jpg'%count)
    count += 1
    plt.savefig(savefile)

▲ 图数字旋转

1.2 模型识别结果

1.2.1 网络识别代码

import sys,os,math,time
sys.path.append("/home/aistudio/external-libraries")
import matplotlib.pyplot as plt
from numpy import *

import paddle
import paddle.fluid as fluid
import cv2
import cv2

imgwidth = 48
imgheight = 48
inputchannel = 1
kernelsize   = 5
targetsize = 10
ftwidth = ((imgwidth-kernelsize+1)//2-kernelsize+1)//2
ftheight = ((imgheight-kernelsize+1)//2-kernelsize+1)//2

class lenet(paddle.nn.Layer):
    def __init__(self, ):
        super(lenet, self).__init__()
        self.conv1 = paddle.nn.Conv2D(in_channels=inputchannel, out_channels=6, kernel_size=kernelsize, stride=1, padding=0)
        self.conv2 = paddle.nn.Conv2D(in_channels=6, out_channels=16, kernel_size=kernelsize, stride=1, padding=0)
        self.mp1    = paddle.nn.MaxPool2D(kernel_size=2, stride=2)
        self.mp2    = paddle.nn.MaxPool2D(kernel_size=2, stride=2)
        self.L1     = paddle.nn.Linear(in_features=ftwidth*ftheight*16, out_features=120)
        self.L2     = paddle.nn.Linear(in_features=120, out_features=86)
        self.L3     = paddle.nn.Linear(in_features=86, out_features=targetsize)

    def forward(self, x):
        x = self.conv1(x)
        x = paddle.nn.functional.relu(x)
        x = self.mp1(x)
        x = self.conv2(x)
        x = paddle.nn.functional.relu(x)
        x = self.mp2(x)
        x = paddle.flatten(x, start_axis=1, stop_axis=-1)
        x = self.L1(x)
        x = paddle.nn.functional.relu(x)
        x = self.L2(x)
        x = paddle.nn.functional.relu(x)
        x = self.L3(x)
        return x

model = lenet()
model.set_state_dict(paddle.load('/home/aistudio/work/seg7model4_1_all.pdparams'))

cutdir = '/home/aistudio/work/7seg/rotate/cutdir'
rotateangle = range(0, 360, 10)

def modelpre(n, degree):
    fn = os.path.join(cutdir, '%d_%03d.BMP'%(n, degree))
    img = cv2.imread(fn)
    data = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    data = data - mean(data)
    stdd = std(data)
    data = data/stdd

    inputs = data[newaxis, :, :]

    preid = model(paddle.to_tensor([inputs], dtype='float32'))

    return preid.numpy()

outdim = []

for r in rotateangle:
    out = modelpre(0, r)
    outdim.append(out[0])

print(outdim)

1.2.2 绘制结果

plt.clf()
plt.figure(figsize=(10,6))
for i in range(10):
    plt.plot([a[i] for a in outdim], label='Out=%d'%i)

plt.xlabel("Angle")
plt.ylabel("Value")
plt.grid(True)
plt.tight_layout()
plt.legend(loc='upper right')
plt.savefig('/home/aistudio/stdout.jpg')
plt.show()