There are a lot of dangerous sloped grounds, includes rock walls, by sides of roads, railways and buildings in JAPAN. In order to prevent accidental disasters, e.g. landslides on the slope surface, the surface movement of the ground must be strictly and continuously monitored. The conventional monitoring method using targets or clinometers on the slopes requires expensive costs, huge manpower and dangerous operation on the slope. This paper shows a new monitoring method for the landslide or slope collapse, requires no targets, using a precision digital camera and image processing algorithms. Serial still images of the slope are firstly taken with the digital camera with a constant time interval. Each image is divided into small block elements. Using a block matching method, all the blocks in the image previously taken are scanned and compared in some search areas in the image subsequently taken, and movement vectors of all the blocks, which means the slope slides or deformations, can be calculated. An image subtraction function and a cross correlation method are utilized as assessment functions in the block matching method. The proposed method is applied to real images of the rock wall collapse and movement vectors are calculated. Small early translational deformation of the rock can be calculated with the method. This result shows a feasibility of the method for monitoring landslides.