在裁剪后的图像上使用pytesseract

Question

I am cropping an image and then using pytesseract on it to read digits from it. However, I have to first save the cropped image and then load it again for this code to work. Is there a way I can avoid doing this and improve efficiency? If I pass cropped directly to the image_to_string it crashes with:

in image_to_string if len(image.split()) == 4: AttributeError: 'numpy.ndarray' object has no attribute 'split'

cropped = image[1044:2000, 100:1025]
cv2.imwrite("thumbnail.png", cropped)
img = Image.open("thumbnail.png")
print(pytesseract.image_to_string(img, config='-psm 6'))
print("--- %s seconds ---" % (time.time() - start_time))

Answer 1

使用Image.fromarray（）

cropped = image[1044:2000, 100:1025] img = Image.fromarray(cropped) print(pytesseract.image_to_string(img, config='-psm 6')) print("--- %s seconds ---" % (time.time() - start_time))