RxSwift 提高性能

这个RxSwift代码运行得很慢，所以你能给我一个如何提高它性能的建议吗？

此功能是获取图像资源并将其映射到重新绘制的图像中

func redrawChapter(pages: [Resource]) -> Single<[UIImage]> {
     return Observable
        .from(pages)
        .flatMap(getImage)
        // MARK: Detect synopsys
        .flatMap(recognizeText)
        // MARK: Translate synopsys
        .concatMap(translate)
        // MARK: Image redraw
        .concatMap(redrawImage)
        .toArray()
}

我在用翠鸟下载图片

private func getImage(resource: Resource) -> Single<UIImage> {
    return Single.create { single in
        let disposables = Disposables.create()
        
        KingfisherManager.shared.retrieveImage(with: resource) { result in
            switch result {
            case .success(let value):
                single(.success(value.image))
            case .failure(let error):
                single(.failure(error))
            }
        }
        return disposables
    }
}

我正在使用Apple Vision来识别文本区域

private func recognizeText(image: UIImage) -> Single<(UIImage, [Synopsis])> {
    return Observable.of(image)
        .flatMap {
            Observable.combineLatest(
                Observable.just($0),
                // MARK: Apple vision to detect text regions
                self.imageProcessor.getRecognizedText(image: $0).asObservable()
            )
        }
        .asSingle()
}

和网络方面的莫亚

private func translate(image: UIImage, synopsys: [Synopsis]) -> Maybe<(UIImage, [Synopsis])>{
    return Observable.from(synopsys)
        .flatMap {
            Observable.combineLatest(
                Observable.just($0.rect),
                self.translator.translate(text: $0.text).asObservable()
            )
        }
        .compactMap {
            return Synopsis(
                text: $0.1,
                rect: $0.0
            )
        }
        .toArray()
        .compactMap {
            return (image, $0)
        }
}

使用UIGraphics渲染器重绘图像

private func redrawImage(
    image: UIImage,
    synopsys: [Synopsis]
) -> Single<UIImage> {
    return Single.create { single in
        
        let format = UIGraphicsImageRendererFormat()
        format.scale = 1
        
        guard let cgImage = image.cgImage else {
            return Disposables.create()
        }
        
        let size = image.size
        
        let bounds = CGRect(
            origin: .zero,
            size: size
        )
        
            let final = UIGraphicsImageRenderer(
                bounds: bounds,
                format: format
            ).image { context in
                
                image.draw(in: bounds)
                
                for syn in synopsys {
                    
                    let backgroundColor = cgImage.averageColorOf(rect: syn.rect)
                    let textColor = backgroundColor.textColor()
                    
                    self.setupLabel(
                        text: syn.text,
                        backgroundColor: backgroundColor,
                        textColor: textColor,
                        bounds: syn.rect,
                        context: context.cgContext
                    )
                }
            }
            single(.success(final))
        
        return Disposables.create()
    }
}

我觉得有另一种方法可以不让如此大量的可观测数据发挥作用，让它发挥作用，也许它会以某种方式提高这种功能的性能.感谢你的帮助.