[python + opencv] 視頻幀圖像去黑邊

視頻幀中經常會出現上下或左右黑邊的情況，爲幀圖像的分析帶來很多麻煩，這裏考慮用掃描k行（或k列）的方法先得到黑邊的寬度，然後直接裁剪圖片。侷限是隻能處理矩形黑邊的的情況。

由於圖像的黑邊一般與圖像的正文有一個過渡帶，很難通過卡閾值的方法去除，所以這裏使用一個shrink值來對去除黑邊後的圖像進行一個縮剪。

#   rmBlackBorder: remove the black borders of one image
#   return: cropped image
def rmBlackBorder(
	src,   # input image
	thres, # threshold for cropping: sum([r,g,b] - [0,0,0](black))
	diff,  # max tolerable difference between black borders on two side
	shrink # number of pixels to shrink after the blackBorders removed
	):
	
	#
	#   remove the black border on both right and left side
	#
	nRow = src.shape[0]
	nCol = src.shape[1]
	left = -1
	right = nCol

	for i in [0,nRow/2,nRow-1]:
		curLeft = -1
		curRight = nCol
		for j in range(0,nCol-1):
			if(sum(list(src[i,j]))<=thres and curLeft == j-1):
				curLeft += 1
		if left == -1:
			left = curLeft
		if curLeft < left:
			left = curLeft

		for j in range(nCol-1,0,-1):
			if(sum(list(src[i,j]))<=thres and curRight == j+1):
				curRight -= 1
		if right == nCol:
			right = curRight
		if curRight > right:
			right = curRight
	
	if min(left,right)>=1 and abs((left+1)-(nCol-right))<=diff and right-left>0 :
		print 'left margin: %d\n' %left
		print 'right margin: %d\n' %right
		
		src = src[0 : nRow-1, left+1+shrink : right-1-shrink, :]
	else:
		src = src

	#
	#   remove the black border on both up and down side
	#
	nRow = src.shape[0]
	nCol = src.shape[1]
	up = -1
	down = nRow
	
	for j in [0,nCol/2,nCol-1]:
		curUp = -1
		curDown = nRow
		for i in range(0,nRow-1):
			if(sum(list(src[i,j]))<=thres and curUp == i-1):
				curUp += 1
		if up == -1:
			up = curUp
		if curUp < up:
			up = curUp

		for i in range(nRow-1,0,-1):
			if(sum(list(src[i,j]))<=thres and curDown == i+1):
				curDown -= 1
		if down == nRow:
			down = curDown
		if curDown > down:
			down = curDown
	
	if min(up,down)>=1 and abs((up+1)-(nRow-down))<=diff and down-up>0:
		print 'up margin: %d\n' %up
		print 'down margin: %d\n' %down
		
		dst = src[up+1+shrink : down-1-shrink, 0 : nCol-1, :]
	else:
		dst = src

	return dst

運行結果如下所示：

原圖：

處理後的結果：