very basic version

main.py

@@ -0,0 +1,81 @@
+import matplotlib.image as mpimg
+import numpy as np
+import matplotlib.pyplot as plt
+import scipy.misc
+
+
+def find_closest_pix(cur_pix,color_set):
+	min_dis = 255*255*3
+	best = color_set[0]
+	for color in color_set:
+		distance = (cur_pix[0]-color[0])**2 + (cur_pix[1]-color[1])**2 + (cur_pix[2]-color[2])**2
+		if min_dis>distance:
+			min_dis = distance
+			best = color
+	return best
+
+def modify_init_pic(name,save_name):
+	img = mpimg.imread(name)
+
+	base_color = [[0,0,0],[41,36,33],[192,192,192],[128,138,135],[112,128,105],
+			  	  [128,128,105],[225,225,225],[250,235,215],[240,255,255],[245,245,245],
+			  	  [255,235,205],[255,248,220],[252,230,201],[255,250,240],[220,220,220],
+			  	  [248,248,255],[240,255,240],[250,240,230],[255,222,173],[253,245,230]]
+	new_img = []
+	for i in range(img.shape[0]):
+		row = []
+		if i%100 == 0:
+			print(i)
+		for j in range(img.shape[1]):
+			cur_pix = img[i][j]
+			new_pix = find_closest_pix(cur_pix,base_color)
+			row.append(new_pix)
+		new_img.append(row)
+	new_img = np.array(new_img,dtype = np.uint8)
+	scipy.misc.imsave(save_name,new_img)
+	print(new_img.shape)
+	print(new_img[0][0])
+	plt.imshow(new_img)
+	plt.show()
+
+
+class KNN:
+	def __init__(self,k,data):
+		self.k = k
+		self.data = data
+
+	def get_new_img(self):
+		new_img = []
+		for i in range(self.data.shape[0]):
+			row = []
+			if i%100 == 0:
+				print(i)
+			for j in range(self.data.shape[1]):
+				neighbors = self.get_k_nearest(i,j)
+				u, indices = np.unique(neighbors, return_inverse=True)
+				most_common = u[np.argmax(np.apply_along_axis(np.bincount,0, indices.reshape(neighbors.shape),None, np.max(indices) + 1), axis=0)].tolist()
+				row.append(most_common)
+			new_img.append(row)
+		new_img = np.array(new_img,dtype = np.uint8)
+		print(new_img.shape)
+		print(new_img[0][0])
+		return new_img
+
+	def get_k_nearest(self,row,col):
+		result = self.data[row][col]
+		for i in range(max(0,row-self.k),min(row+self.k,self.data.shape[0])):
+			for j in range(max(0,col-self.k),min(col+self.k,self.data.shape[1])):
+				if i!=row and j!=col:
+					result = np.vstack((result,self.data[i][j]))
+		return result
+
+if __name__ == '__main__':
+	#modify_init_pic("init.jpg","3.jpg")
+	img = mpimg.imread('3.jpg')
+	k = 4
+	knn = KNN(k,img)
+	new_img = knn.get_new_img()
+	plt.imshow(new_img)
+	plt.show()
+
+