Multi-lens image stitching device and method

Abstract

The invention provides a multi-lens image stitching device and method, and relates to the field of integrated circuit chip design. The method comprises the following steps that: a front processing module executes multi-image ISP processing, distortion correction and geometric mapping transformation on a plurality of original images to obtain a plurality of real-scene images; a multi-dimensional visual processing module performs object detection and recognition on the real-scene image to obtain a multi-dimensional structured image; a multi-image stitching module combines the real-scene image and the multi-dimensional structured image in various combination ways to obtain stitched combination images and stitching combination parameters; and a coding output module arranges and codes the stitching combination images, and outputs the coded image and stitching combination parameters. All logic modules for image processing are integrated in one integrated circuit chip, thereby improving the efficiency of multi-image stitching combination processing and the flexibility of dynamic configuration, lowering the difficulty in software development of back-end multi-image stitching display, lowering the development cost and power consumption, and reducing the product size.

Description

technical field [0001] The invention relates to the field of integrated circuit chip design, in particular to a multi-lens image splicing device and method. Background technique [0002] The application of multi-camera image splicing and display has become very popular. The mainstream method is to use a dedicated image splicing chip to collect multi-camera videos, arrange and combine them, and then display them on the screen through HDMI and other video interfaces. [0003] Currently there are two main schemes, one is ASIC / FPGA (Application Specific Integrated Circuit) / (Field-Programmable Gate Array, Field Programmable Gate Array) scheme, which receives multiple video signals through an external communication interface, and the video signal After decompression or AD sampling, it is converted into a digital signal, and then the video is synthesized by hardware, such as the video splitter on the market. The second is the software synthesis scheme, which needs to use the image...

AI Technical Summary

Problems solved by technology

[0004] However, the current existing solutions have the following disadvantages: since the camera is a single-lens device, after multiple cameras perform video encoding, they are integrated through a router, and then the video splitter or combiner performs large-screen image splicing and display

Its disadvantage is that it cannot realize the operation of multi-sensor image acquisition, splicing and integration, and requires centralized acquisition of multi-channel camera images, multi-channel wiring, high overall cost and high power consumption

[0005] The software synthesis scheme requires the cooperation of computer and software, which is large in size and high in cost. When the number of videos to be processed is large, it takes up a lot of bandwidth and system resources, and its real-time performance is easily affected by CPU (Central Processing Unit, central processing unit) etc. factors

[0006] If the existing solution is splicing and combination at the pixel level, it will inevitably involve video data decompression, etc., which will consume a lot of hardware resources. As the number of channels increases, the resources consumed will also increase by multiples; If the compressed data stream is spliced, the image cannot be rescaled, and the flexibility is lost

Especially for multi-camera signal collection and display in closed environments such as homes, small shops, and vehicles, it is necessary to purchase and integrate a complete set of connected split-screen devices, which greatly increases installation costs and deployment difficulties.

Images